Multi-Level Regulation of Opioid-Induced Respiratory Depression

Abstract

Opioids depress minute ventilation primarily by reducing respiratory rate. This results from direct effects on the preBötzinger Complex as well as from depression of the Parabrachial/Kölliker-Fuse Complex, which provides excitatory drive to preBötzinger Complex neurons mediating respiratory phase-switch. Opioids also depress awake drive from the forebrain and chemodrive.

Introduction

Respiration is an automatic process that ensures adequate oxygen uptake and carbon dioxide removal from the body. The primary drive to breathe is derived from chemoreception, and respiratory phase duration and pattern are influenced by feedback from lung and airway receptors (110). Automatic breathing is generated in the brain stem and continues during sleep as well as under sedation (FIGURE 1). In the awake state, respiration is heavily influenced by the cortico-limbic system of the forebrain (review in Ref. 44), and respiratory activity changes promptly with arousal (70), emotions, and the level of physical activity. Volitional control allows for breath-holds, hyperventilation, coughing, and complex motor functions like vocalization (72). Loss of “awake drive” during sleep results in a decrease in respiratory minute ventilation (49) and a reduced hypoxic ventilatory response (11, 91, 118).

FIGURE 1.

Excitatory connections within the respiratory control center and opioid effects

Tonic chemodrive (green solid arrows) is the main excitatory drive to the medullary rhythm generator. A large component of chemodrive is routed through the Parabrachial Nucleus/Kölliker-Fuse Complex to phase-switching neurons in the preBötzinger Complex and determines respiratory rate (dark blue solid arrow). Phasic inputs from the preBötzinger Complex activate inspiratory (I) and expiratory (E) premotor and motoneurons (blue dotted arrows). Respiratory motor output (tidal volume) depends on direct projections from the retrotrapezoid nucleus to preBötzinger Complex neurons, premotor neurons, and motoneurons. The cortico-limbic system contributes tonic drive to the medullary rhythm generator (light blue solid arrows). Not shown: direct projections from the motor cortex to phrenic motoneurons as a pathway to override automatic rhythm; projections from the hypothalamus and cerebellum to the medullary raphe, which may contribute to the state-dependency of respiratory activity. Opioid effects: Parabrachial Nucleus/Kölliker-Fuse Complex activity was depressed in all studies and at all opioid concentrations. Opioid-induced depression was also shown for the preBötzinger Complex, the nucleus of the solitary tract, and the medullary raphe. Opioid-induced sedation suggests depression of forebrain inputs. Premotor neurons are only directly depressed at very high opioid concentrations. Inset: neuronal subtypes constituting the core of respiratory rhythm generation. Through mutual excitation, network activity of pre-inspiratory neurons (pre-I) in the preBötzinger Complex results in activation of early inspiratory (early-I) neurons (green arrows). Activity of these neurons is terminated through inhibition by post-inspiratory neurons (post-I), which themselves are inhibited by early-I neurons during the inspiratory phase (red circles). Phasic excitation is relayed to inspiratory (I) and expiratory (E) premotor neurons.

Opioid-induced respiratory depression (OIRD) is an important problem in the perioperative period (84) and increasingly in the community (170). The respiratory depressant effect is dose-dependent, and the magnitude of respiratory depression correlates with the level of sedation and analgesia (29, 92). In the clinical setting, opioids can cause mild sedation, hypoventilation, an increase in Pco₂ above 50 Torr, and a decrease in oxygen saturation already with standard analgesic doses (20, 53, 74, 116a, 165). At that stage, patients can be prompted to breathe by voice command or touch. With deeper sedation, painful stimuli are still able to arouse the patient and elicit respiration. Pco₂ can exceed 60 Torr (20) (53, 74, 165). Very high opioid doses depress respiration to a degree where neither severe hypoxia nor hypercapnia nor pain will be sufficient to generate respiratory efforts (90). These “overdoses” require artificial ventilation or treatment with the opioid antagonist naloxone (20). The gradual decline in minute ventilation and the concomitant increase in sedation suggest that OIRD involves multiple areas in the forebrain and brain stem.

This review describes the mechanism of respiratory rhythm generation in the brain stem, how excitatory drive from chemoreceptors and the forebrain contributes to minute ventilation, and how these areas are affected by opioids. Since opioids mostly depress respiratory rate, and apnea results from an arrest of the respiratory cycle in the expiratory phase rather than from a severe decrease in respiratory tidal volume (48, 83, 97, 127, 137), we particularly highlight the mechanism of respiratory rate control and how it is depressed by opioids. We focus on in vivo studies that examine respiratory mechanisms at physiological levels of respiratory drive, and we present data obtained in human subjects wherever available.

Organization of the Respiratory System

Respiratory Rhythm Generation

Respiratory rhythm originates in the ventrolateral medulla where the neuronal network of the preBötzinger Complex and Bötzinger Complex converts tonic excitatory respiratory drive into a distinct inspiratory and expiratory phase (FIGURE 1). Phasic respiratory activity is relayed via premotor neurons in the caudal medulla to motoneurons in the spinal cord. Phrenic motoneurons excite the main inspiratory muscle, the diaphragm, whereas thoraco-abdominal motoneurons phasically excite abdominal muscle activity during active expiration (34, 88, 130). The neurons of the preBötzinger Complex, the Bötzinger Complex, and the adjacent parafacial respiratory group can be classified by location, discharge pattern, neurotransmitters, genetics, and connectivity (see reviews in Refs. 34, 56, 130, 132). There is significant topical overlap between functionally distinct neuronal populations (5, 140, 172).

Respiratory rhythm generation relies on the activity of pre-inspiratory neurons to start the inspiratory phase (“inspiratory on-switch”; FIGURE 1, INSET). Synchronized activity of a sufficient number of pre-inspiratory neurons (76) activates other inspiratory neurons in the preBötzinger Complex to generate a full inspiratory cycle (2, 34, 50, 104). The level of excitability of preBötzinger Complex inspiratory neurons is modulated by inhibitory inputs, which affect the rate of depolarization and the duration of the refractory period (2, 4) (FIGURE 2). Inspiration is terminated when inspiratory neurons are inhibited by post-inspiratory neurons (“inspiratory off-switch”) (27, 45, 104). Research is ongoing to elicit whether respiratory rate is determined mostly by the depolarization rate of pre-inspiratory and inspiratory neurons in the preBötzinger Complex (71) or whether it equally depends on the activity of expiratory neurons in the Bötzinger Complex (7, 104, 132) that inhibit inspiratory neurons (71). In addition to neurons contributing to respiratory phase switching, the preBötzinger and Bötzinger Complexes also contain inspiratory and expiratory neurons whose discharge prolongs phase duration and contributes to inspiratory and expiratory motor output (tidal volume; FIGURE 1) (88, 141).

FIGURE 2.

Model of the components determining the membrane potential and illustrating the effects of changes in excitatory and/or inhibitory inputs

A: model of the components determining the membrane potential (V_m) of a preBötzinger Complex phase-switching neuron: ${\text{V}}_{\text{m}}=\left({\text{g}}_{\text{exc}}·{\text{V}}_{\text{E}}\right)+\left({\text{g}}_{\text{inh}}·{\text{V}}_{\text{I}}\right)/\left({\text{g}}_{\text{exc}}+{\text{g}}_{\text{inh}}+{\text{g}}_{\text{leak}}\right)$, where g_exc is excitatory conductance, V_E is equilibrium potential for excitatory currents, g_inh is inhibitory conductance, V_I is equilibrium potential for inhibitory currents, and g_leak is leak conductance, e.g., GIRK channels (see text for details). B and C: functional timer model to illustrate the effects of changes in excitatory and/or inhibitory inputs on the neuronal membrane trajectory (see Refs. 28, 115, 171–173). B: the sum of all excitatory and inhibitory inputs (σ) is gated to a leaky integrator (LI) at time, t = 0 s. The magnitude of σ determines the rate of the exponential rise of V_m to a threshold (Thr). Crossing the threshold resets the leaky integrator via a comparator (COM). The time to crossing the threshold determines the phase duration. C: graphic illustration of the timing operation using the example of rhythmogenic pre-inspiratory neurons. Neuronal discharge in pre-inspiratory neurons begins when Thr is reached and results in inspiratory on-switch. Inspiratory on-switch terminates the expiratory phase, i.e., the time to Thr for pre-inspiratory neurons determines expiratory duration (T_E). Due to the nonlinear nature of this mechanism, increases and decreases in σ of the same magnitude cause strikingly different changes in phase duration. Shown are three examples for σ (upper) and the corresponding leaky integrator outputs LI (σ) (lower). Setting σ = 1.0 as a baseline reference results in a duration of T_E2 = 1.6 s (black lines). Increasing σ by 40% results in T_E1 = 1 s (red), whereas decreasing σ by 40% results in T_E3 = 5 s (blue). Physiological examples for an increase in σ could be an increase in neuronal activity in the parabrachial nucleus (PBN)/Kölliker-Fuse nucleus (137) or an increase in inhibitory activity during the preceding inspiratory phase, which shortens inspiratory duration and presumably shortens the post-inspiratory refractory period (3). The latter could be due to vagal pulmonary stretch receptor input during lung inflation or increased activity of preBötzinger Complex inhibitory neurons during the inspiratory phase (3). Decreases in σ could be due to a reduction in PBN activity via inhibition of PBN neurons by opioids, by increases in pulmonary stretch receptor activity during the expiratory phase (173), or by increases in preBötzinger Complex GABAergic/glycinergic neuronal activity during the expiratory phase (3).

Using comparative cytoarchitecture and immunohistochemistry, Schwarzacher et al. identified the equivalent of the preBötzinger Complex in the human ventrolateral medulla (139). Functional MRI studies showed inspiratory activity in this area that alternated with expiratory activity in the adjacent caudal ventrolateral pons/parafacial group (63). Interestingly, preBötzinger Complex activity was increased during loaded inspiration in healthy controls, whereas patients with chronic obstructive pulmonary disease, which is associated with impaired airflow during expiration, showed increased activation of the parafacial group (63).

Respiratory Drive

The term “respiratory drive” is used broadly in respiratory control to describe excitatory inputs to brain stem areas that increase respiratory activity. Respiratory chemodrive summarizes the tonic excitatory inputs to the medullary rhythm generator that result from activation of chemoreceptive brain stem areas and the carotid body (56) by Pco₂ and hypoxia (FIGURE 1). Hypoxia (12) and hypercapnia (30) cause large increases in minute ventilation. In humans, this was associated with increased activity in the carotid body and the nucleus of the solitary tract (101, 126). Short episodes of mild hypercapnia also increased activity in the thalamic nuclei, the pontine raphe, the Parabrachial Nucleus/Kölliker-Fuse Complex, and the locus coeruleus (126). No additional increase was observed once Pco₂ exceeded ~65 Torr (81).

A large component of respiratory chemodrive originates in the retrotrapezoid nucleus, which contains chemosensitive neurons (111, 144, 156), and also integrates the peripheral chemoreceptive inputs from the carotid body (55). Individuals with a genetic lack of chemoreceptive retrotrapezoid nucleus neurons (congenital central hypoventilation syndrome) suffer from severe hypoventilation during sleep (129, 161). The retrotrapezoid nucleus projects glutamatergic excitatory inputs to the Bötzinger Complex, preBötzinger Complex, Parabrachial Nucleus/Kölliker-Fuse Complex, and premotor neurons, and phrenic motoneurons (14, 17). Glutamate receptor antagonism in the preBötzinger Complex results in short, irregular, low-amplitude breaths and ultimately apnea (25, 113), highlighting the importance of glutamatergic drive for respiratory function (FIGURE 3A).

FIGURE 3.

Phrenic neurogram tracings obtained in adult decerebrate rabbits in vivo with Pco₂ constant

A: antagonism of glutamate receptor function in the preBötzinger Complex (preBötC) through local microinjection of the AMPA receptor antagonist NBQX (blue) and the NMDA receptor antagonist AP5 (red) caused tachypnea with decreased peak phrenic amplitude, followed by apnea (25). B: antagonism of glutamate receptor function in the parabrachial nucleus (PBN) and Kölliker-Fuse nucleus (KF) through local microinjection of NBQX (blue) and AP5 (red) caused severe bradypnea, whereas peak phrenic amplitude was only little decreased. Complete apnea was never observed (116). Please note the different time scales. C: repeated intravenous boluses of the mu-opioid agonist remifentanil (Remi; 1 mcg · kg^–1 $·$ bolus^–1) caused increasing bradypnea due to prolongation of inspiratory and expiratory phase duration. The fourth bolus resulted in apnea (arrow). Remifentanil also substantially depressed peak phrenic activity (123), suggesting an effect on both the PBN/KF and the preBötC.

A second chemosensitive system consists of the neurons of the medullary raphe (see reviews in Refs. 26, 66). Raphe neurons form serotonergic synapses, at times colocalized with substance P, with neurons in the preBötzinger Complex and the pontine respiratory group (128). Both neurotransmitters cause a decrease in membrane potential of preBötzinger Complex neurons and thus modulate their level of excitability (128). Input from the raphe obscurus modulated the pH response of retrotrapezoid nucleus neurons (162). The raphe obscurus receives multiple synaptic inputs from suprapontine areas (below) and preBötzinger Complex inspiratory neurons (128), suggesting that the raphe contributes to state-dependent modulation of respiratory activity (128).

Additional respiratory drive originates in the cortico-limbic system. In humans, MRI diffusion tractography established prominent connections between the hippocampus and the raphe, locus coeruleus, and nucleus paragigantocellularis lateralis (42), the human equivalent of the preBötzinger Complex (139). This connection may be responsible for increasing respiration as part of the “fight-or-flight” response. Hippocampus and amygdala appear to mediate the “urge to breathe” during hypercarbia-stimulated or inspiratory-resistive loaded breathing (61, 126). In anesthetized rats, stimulation of hypothalamic nuclei increased respiratory rate and tidal volume (Ref. 158 and review in Ref. 77). Many of these efferents are relayed by the Parabrachial Nucleus/Kölliker-Fuse Complex (22), but some directly project to the spinal cord (21). The hypothalamus also sends orexinergic projections to the raphe magnus, raphe obscurus, and retrotrapezoid nucleus, which do not affect baseline respiration but enhance CO₂ sensitivity during the nocturnal active phase in rats (review in Ref. 114). Direct connections between the motor cortex and phrenic motoneurons were shown histologically (133) and functionally (8) in cats. In humans, functional MRI showed an activation of the motor cortex with inspiratory loading (Ref. 166, and see review in Ref. 44). Broadly, the activity of these areas contributes tonic “awake drive” to the medullary respiratory center (105, 117). Additional awake drive is produced by the lateral reticular formation of the medulla (117), emphasizing the state dependency of respiratory drive.

The periaqueductal gray lies within the tegmentum of the midbrain, at the juncture between the forebrain and the brain stem. The periaqueductal gray provides an important connection for sensory feedback from the nociceptive areas of the spinal cord and lung mechanoreceptors relaying work of breathing (23, 24) toward the thalamus (95) and, in the opposite direction, for drive from the prefrontal, premotor, motor, and cingular cortex (33, 96) to the locus coeruleus, lateral PBN, nucleus ambiguous, and raphe magnus and pallidus (94). It thus contributes to the integration of the respiratory pattern, vocalization, and upper airway maneuvers with changing emotions and behaviors (Ref. 151, and reviewed in Ref. 153). Stimulation of a subarea of the periaqueductal gray resulted in tachypnea (152, 153). This response was significantly reduced by inhibition of the lateral PBN (62), suggesting that respiratory drive coming from the periaqueductal gray was relayed by the PBN. In cats, activity of the cerebellar fastigial nucleus did not alter baseline respiratory rate but increased the response to severe hypercapnia and hypoxia (163, 164). Anatomical connections suggest that this effect is also mediated by the PBN (154).

Multilevel Opioid Effects

The preBötzinger Complex and Parabrachial Nucleus/Kölliker-Fuse Complex Are Prime Targets for Drugs That Depress Respiratory Rate

The concomitant increase of respiratory rate and tidal volume with hypoxia and hypercapnia may suggest that both parameters are tightly linked. However, the parameters can be “uncoupled” by reducing Parabrachial Nucleus/Kölliker-Fuse Complex activity (39, 41, 47, 89, 116). The Parabrachial Nucleus/Kölliker-Fuse Complex is an important relay station for tonic excitatory inputs from the retrotrapezoid nucleus (142) and the medullary raphe, as well as from the forebrain, the periaqueductal gray, the cerebellum, and the ascending nociceptive inputs (73) to the medullary rhythm generator. In decerebrate rabbits, blockage of glutamatergic inputs to the Parabrachial Nucleus/Kölliker-Fuse Complex depressed respiratory rate by >90% but decreased peak phrenic activity only by <20% (116) (FIGURE 3B). Subsequent exposure to hypoxic hypercapnia during persistent glutamatergic block increased respiratory rate only to 25% of control, whereas peak phrenic activity increased to 160% of control.

Considering that the respiratory rhythm is generated in the preBötzinger Complex, a possible explanation for this differential effect is that excitatory drive to the preBötzinger Complex is not nonspecific, as depicted in some models (54, 132), but that different brain stem areas specifically excite certain neuronal subpopulations (7, 116). Histological and functional studies have demonstrated excitatory projections from the retrotrapezoid nucleus to the Parabrachial Nucleus/Kölliker-Fuse Complex (14, 142) and preBötzinger Complex (14) and also from the Parabrachial Nucleus/Kölliker-Fuse Complex to the preBötzinger Complex (51, 172). The importance of Parabrachial Nucleus/Kölliker-Fuse Complex activity for respiratory rate suggests that the majority of excitatory drive to phase-switching neurons in the preBötzinger Complex/Bötzinger Complex is relayed through the Parabrachial Nucleus/Kölliker-Fuse Complex. On the other hand, inputs that determine respiratory tidal volume likely project directly from the retrotrapezoid nucleus to neurons in the medullary rhythm generator as well as premotor neurons and motoneurons in the spinal cord (FIGURE 1).

As described above, inspiratory on-switch depends on the depolarization rate of pre-inspiratory and inspiratory preBötzinger Complex neurons (4, 5, 18, 19, 171, 172) (FIGURE 2C), whereas inspiratory off-switch is likely regulated by the depolarization rate of post-inspiratory neurons (7, 41, 116). FIGURE 2C illustrates that, when the sum of inputs (σ) is large, i.e., excitatory inputs (${\text{g}}_{\text{exc}}·{\text{V}}_{\text{E}}$) to a phase-switching neuron are much higher than inhibitory inputs (${\text{g}}_{\text{inh}}·{\text{V}}_{\text{I}}$), the membrane potential quickly reaches its discharge threshold, and the duration of the preceding respiratory phase (T) is short. In this case, a small decrease in excitatory drive causes only a small increase in phase duration. In contrast, when the sum of inputs is small, time to phase-switch is prolonged. Under these conditions, a decrease in excitatory drive from the Parabrachial Nucleus/Kölliker-Fuse Complex to pre-inspiratory neurons or direct inhibition of pre-inspiratory neurons would substantially prolong the expiratory phase. Furthermore, a small additional decrease in excitatory drive or increase in inhibition would result in a very long expiratory phase (116) and potentially apnea. This explains the observation that respiratory rate variation is greater when excitatory drive to the rhythm generator is low (19).

Opioids hyperpolarized respiratory neurons through a mu-opioid-receptor-coupled G-protein-gated inwardly rectifying potassium (GIRK) conductance (g_leak; FIGURE 2A) in the preBötzinger Complex and Kölliker-Fuse nucleus in vitro (85, 108, 160). Opioids thus have the potential to slow depolarization of phase-switching neurons in the preBötzinger Complex either through a direct, inhibitory effect on these neurons or through inhibition of Parabrachial Nucleus/Kölliker-Fuse Complex neurons, which reduces the excitatory drive to the preBötzinger Complex.

Opioid Effects on Respiratory Rate and Motor Output in vivo

Multiple studies point to a direct opioid effect on preBötzinger Complex neurons. In mice in vivo, opioids activated a GIRK channel, and block of GIRK channels in the preBötzinger Complex reduced the respiratory depression from systemic mu-opioid agonists (108). Microinfusion of the opioid antagonist naloxone into the preBötzinger Complex completely prevented the 25% respiratory rate depression from small doses of IV fentanyl in rats (107). In contrast, in in vivo dogs (112) and rabbits (145), naloxone injection into the bilateral preBötzinger Complex did not reverse a 50% respiratory rate depression caused by an intravenous remifentanil infusion; further analysis suggested that the opioid effect on expiratory duration was mediated mostly outside the preBötzinger Complex (145). Interestingly, in vitro bath application of mu-opioid receptor agonists inhibited inspiratory preBötzinger Complex neurons (102, 157) but did not affect pre-inspiratory neurons (102) or expiratory neurons (157).

In the in vivo rabbit preparation, naloxone injection into the parabrachial nucleus partially reversed respiratory rate depression from an intravenous remifentanil infusion from 50% to 20% (103). In the in vivo decerebrate dog preparation, microinjected naloxone into the parabrachial nucleus produced a full reversal of remifentanil-induced bradypnea/apnea (127). In the in situ rat model, injection of the opioid antagonist CTAP into the bilateral Kölliker-Fuse nucleus prevented apnea from systemic fentanyl infusion (137). In rabbits and rats, respiratory rate depression was not fully reversed, with local microinjections of opioid antagonists in the parabrachial nucleus or Kölliker-Fuse nucleus. The quantitative differences in the naloxone effects may be due to species differences.

In freely behaving mice, mu-opioid receptor deletion in the bilateral Kölliker-Fuse nucleus reduced the respiratory rate depression from small (10 mg/kg), intermediate (30 mg/kg), and high (100 mg/kg) morphine doses by ~20% at each dose. Mu-opioid receptor deletion in the bilateral preBötzinger Complex also reduced respiratory rate depression after 10 mg/kg morphine; however, there was no effect at higher morphine doses (160). In a similar model, mu-opioid receptor deletion in the preBötzinger Complex attenuated the respiratory rate depression from 20 mg/kg morphine from 50% to 70% of control rate (3). Additional attenuation of the morphine effect from subsequent mu-opioid receptor deletion in the second area was not statistically significant; however, the study may have been underpowered to show the effect (3) (FIGURE 4). Taken together, these studies show that opioid-induced respiratory rate depression results in a large degree from a combination of direct depression of the preBötzinger Complex and a decrease in excitatory drive from the Parabrachial Nucleus/Kölliker-Fuse Complex to the preBötzinger Complex. The depression of each area appears to be dose-dependent and species-dependent.

FIGURE 4.

Critical importance of the Parabrachial Nucleus/Kölliker-Fuse complex and preBötzinger Complex for opioid-induced respiratory depression

A–C: μ-opioid receptors (Oprm1) were selectively deleted in Oprm1 ^f/f mice through Cre-virus injection into the preBötC and PBN/KF in this (cohort 1) or reverse order (cohort 2) with 4–5 wk between injections and plethysmography recordings (data from A–C from Ref 3, and used with permission from eLife). A: probability density function plot of the respiratory rate for a representative animal from cohorts 1 and 2 after intraperitoneal morphine (20 mg/kg) or saline before and after Oprm1 deletion in the PBN/KF and preBötC. B: pooled data showed that morphine-induced respiratory rate depression was significantly attenuated after Oprm1 deletion in the preBötC. Oprm1 deletion in both areas reduced morphine-induced respiratory rate depression from ~50% to ~30%. C: plethysmography recordings in mice show that Oprm1 deletion in the preBötC + PBN/KF prevented respiratory rate depression from a very high intraperitoneal fentanyl dose (150 mg/kg; right) that usually caused lethal apnea before Oprm1 deletion (left). D: consistent with the murine studies, phrenic neurogram tracings obtained in an adult decerebrate rabbit in vivo show that sequential microinjections of the opioid antagonist naloxone into the bilateral PBN + KF and bilateral preBötC reversed the respiratory rate depression from intravenous remifentanil infusion (see Ref. 123).

A recent study in freely behaving rats demonstrated the existence of a non-opioid-sensitive, chemo-insensitive respiratory rhythm that developed after very high opioid doses (300 µg/kg fentanyl) (59). Respiratory activity appeared to be driven primarily by expiratory abdominal efforts (59). In a different rat model, phasic expiratory activity was observed even after inspiratory activity was suppressed by fentanyl (102). Although the severe hypoxia and acidosis associated with the high fentanyl dose (58) makes it unlikely that humans would survive to develop such rhythm, the study raises the interesting question of whether opioids also inhibit inhibitory pathways that normally suppress phasic respiratory activity.

Low doses of systemic opioids that decreased respiratory rate did not directly depress bulbospinal premotor neurons, although mu-opioid receptors were present on these neurons (83, 149). Near-apneic fentanyl doses hyperpolarized premotor neurons, suggesting that high doses directly depress neuronal activity (57, 83). Remifentanil infusion depressed phrenic nerve amplitude more than peak premotor neuronal activity, suggesting an additional, direct depressant effect of opioids on phrenic motoneurons (149). To date, no studies have investigated the effects of opioids on brain stem mechanisms in humans.

Chronic opioid use leads to desensitization and cellular tolerance (86), resulting in increased opioid requirements to maintain the same level of intended analgesia or euphoria. At the same time, higher opioid doses may be tolerated before respiratory depression occurs. In freely behaving mice, chronic exposure to morphine resulted in a decrease in respiratory depression from systemic bolus doses of morphine and fentanyl (64). However, respiratory depression was not reduced at high fentanyl doses, suggesting only a limited cross tolerance between morphine and fentanyl (64). Multiple clinical studies have highlighted that the risk of (fatal) overdose in patient populations was increased at higher total opioid doses, suggesting that chronic opioid use did not eliminate the dose-dependency of respiratory depression (reviewed in Ref. 38). There is no evidence that the locations or network mechanism of OIRD differ between opioid-naive and opioid-tolerant individuals.

Opioid Effects on Respiratory Drive in vivo

In human volunteers, analgesic doses of morphine decreased the hypoxic ventilatory response (12) (FIGURE 5A) as well as the ventilatory response to hypercarbia (30) (FIGURE 5B). This was possibly due to an effect on the nucleus of the solitary tract: In anesthetized rats, injection of the mu-opioid antagonist CTAP into the commissural subnucleus of the nucleus of the solitary tract almost completely reversed the depression of the hypoxic ventilatory response caused by intravenous opioids (169). In the same model, injection of CTAP into the caudal medullary raphe partially reversed the depression of the hypercapnic ventilatory response from systemic opioids (168). In contrast, in in vivo rats, morphine doses that caused apnea did not decrease discharge frequency or CO₂ response in chemosensitive retrotrapezoid nucleus neurons (111).

FIGURE 5.

Contributions of chemodrive and “awake drive” to minute ventilation and opioid effects

A: hypoxic ventilatory response curves in a chloralose-urethane anesthetized cat during control (solid square) and after administration of 0.15 mg/kg IV morphine. Morphine decreased minute ventilation during hyperoxia but did not change the increase in minute ventilation with hypoxia (see Ref. 12). B: mean ventilatory response to increasing inspiratory carbon dioxide concentrations obtained in 12 male (square) and 12 female (circle) human volunteers during control (filled) and after 0.1 mg/kg IV morphine, followed by 0.03 mg · kg^–1 · h^–1 (open). The continuous lines are the linear regression lines through the data points, and broken lines are extrapolated to the apneic threshold. In men and women, morphine decreased minute ventilation differently: in men by increasing the apneic threshold but in women by decreasing carbon dioxide sensitivity (see Ref. 30). C–E: association between respiratory depression from analgesic doses of morphine and loss of “awake drive” per electroencephalogram in pediatric patients. C: after 0.185 mg/kg morphine, patient 2 presented substantial respiratory rate depression associated with a decrease in β₁ power. D: in patient 3, a similar dose (0.178 mg/kg) did not reduce β₁ power or cause notable respiratory depression. E: in 10 patients, the severity of respiratory rate depression correlated with the intensity of the reduction in β₁ power (R = 0.715, P = 0.02) (105).

Opioids also dose-dependently cause sedation, and this effect reduces or eliminates excitatory forebrain inputs to the brain stem (16, 29, 92, 135) (FIGURE 1). Functional MRI in human volunteers who received sedative doses of remifentanil showed decreased activity in the prefrontal cortex, anterior cingulate, thalamus, subthalamic nucleus, cerebellum, and periaqueductal gray (125), i.e., in areas that mediate pain and other unpleasant sensations and that contain a high concentration of opioid receptors (9). This may explain why already small opioid doses eliminate the “urge to breathe” during breathholds or with inspiratory CO₂ challenges (99, 125). This effect can be clinically exploited to reduce the sensation of air hunger in patients with heart failure (43). It may also explain why the severe hypoxia and hypercapnia after opioid overdoses do not cause patient distress (46).

In anesthetized rats, intravenous morphine inhibited acetylcholine release in the prefrontal cortex and decreased arousal (121). Direct fentanyl injection into hypothalamic subnuclei led to a decrease in respiratory rate for >20 min (159). The importance of the cortical arousal state was shown in patients who received morphine for postoperative pain control. Respiratory rate was decreased on average by 8%, and the decrease correlated with the decrease in beta-1 power in the electroencephalogram (105) (FIGURE 5, C–E). In freely behaving rats, 100 µg/kg fentanyl—approximately equivalent to a strong analgesic dose of 5 µg/kg in humans—caused significant sedation and reduced alpha and beta-2 power in the EEG recording (106). The increase in theta power correlated with respiratory rate depression (106). Interestingly, opioids did not affect the motor cortex (125), i.e., as long as patients remain awake, they are able to breathe on command (84).

The role of respiratory subareas in OIRD depends on the relative contribution of these areas to respiratory activity as well as on their sensitivity to different opioid concentrations. Until now, in vivo studies have not differentiated between these two factors. In addition, more research is needed to determine whether the opioid effect on any individual area is modified by changed inputs from other areas. No study to date has assessed the relative contributions to OIRD of all factors and all affected areas, i.e., chemodrive, awake drive, the Parabrachial Nucleus/Kölliker-Fuse Complex and the preBötzinger Complex, in the same model.

Future Research

Sedative Drugs

Clinically used sedatives like the alpha-2 antagonist dexmedetomidine (78a) and the GABA_A-receptor agonist midazolam cause limited respiratory depression even at high doses, whereas the NMDA-receptor antagonist ketamine and the GABA_A agonist propofol can cause severe respiratory depression and apnea (see review in Ref. 150). All of these drugs enhance the respiratory depressant effect of opioids (67, 122). Benzodiazepines, for example, are frequently found in overdose victims in the community (75). Alpha-2 receptors have not been identified in the preBötzinger Complex or Parabrachial Nucleus/Kölliker-Fuse Complex, suggesting that any enhancement of OIRD may be due to removal of “awake drive.” NMDA and GABA_A receptors have been located in all respiratory-related brain stem areas, providing potential targets for sedative agents (25, 27, 36, 37, 41, 79, 98, 113, 115, 158). We hypothesize that sedatives affect respiratory rate through the same mechanism as opioids, i.e., by decreasing neuronal excitability of phase-switching neurons in the preBötzinger Complex (FIGURE 2A). The main excitatory inputs to preBötzinger Complex neurons are glutamate receptor mediated (25, 113), whereas inhibitory inputs are GABA_A or glycine receptor mediated (4, 27). Sedatives that block glutamate-receptor function or enhance GABA_A-receptor function thus have the potential to decrease respiratory rate, similarly to opioids, either through direct depression of preBötzinger Complex neurons or through depression of Parabrachial Nucleus/Kölliker-Fuse Complex activity, which lowers glutamatergic drive to the preBötzinger Complex. This would explain how sedative doses that cause only small decreases in membrane excitability and thus respiratory rate when the drug is given by itself can result in severe respiratory slowing or apnea when added after neuronal excitability is already decreased by systemic opioids (FIGURE 2C). In addition, sedatives may depress the activity of other preBötzinger Complex neurons, premotor neurons, and phrenic motoneurons, and thus contribute to a decrease in tidal volume. The exact locations mediating respiratory depression from sedatives have been studied in far less detail than opioids (150), and, just as for opioids, the effects on individual areas may depend on the drug dose (Table 1).

TABLE 1.

Studies of opioid and sedative effects in individual respiratory-related areas

Brain Region	Mu-Opioid Receptor	NMDA Receptor	GABAA Receptor
Carotid body	12, 78	65	65, 167
Nucleus tractus solitarii	30, 169	124
Retrotrapezoid nucleus	111	109	87, 155
Medullary raphe	168	10	87
Forebrain	105, 121, 125	93, 120	1, 60, 69, 119
Periaqueductal gray			138
Parabrachial Nucleus/Kölliker-Fuse Complex	3, 85, 103, 137, 160	40, 115	32, 35, 142
Pre-Bötzinger Complex	3, 107, 112, 145, 160	25, 113	15, 27, 98
Premotor neurons	57, 83, 149	80, 134	36, 100, 146, 148
Phrenic motoneurons	68, 149	100	100

Numbers correspond to reference numbers. Bold: where available, we quote “clinically relevant studies,” i.e., in vivo studies that record respiratory neuronal or global respiratory output during systemic drug application and localized antagonist injection or receptor deletion. Italic: if such studies are not available, we present in vivo studies recording respiratory-related neurons or global output during localized application of the sedative agent or the respective receptor agonist or antagonist at clinically relevant or higher concentrations; alternatively, these are descriptive studies using fMRI or EEG activity in humans. Roman: if no in vivo studies are available, we present in vitro studies using localized application of the receptor agonist/antagonist or indirect evidence. Sedatives included are NMDA receptor antagonists (e.g., ketamine) and GABA_A receptor agonists (e.g., Propofol and midazolam). Although clinically relevant opioid concentrations have been studied in vivo in many areas, this research still needs to be performed for clinically used sedatives.

Respiratory Stimulants

We have described that opioids depress respiratory activity through effects on multiple areas of the central nervous system. This is of practical relevance for the development of pharmacological agents designed to counteract OIRD (see detailed review in Ref. 31). For example, opioid-induced depression of the Parabrachial Nucleus/Kölliker-Fuse Complex significantly reduces excitatory drive to the preBötzinger Complex. This likely limits the respiratory-stimulating effects of receptor agonists that specifically stimulate neurons in the preBötzinger Complex (82, 97). Similarly, opioids depress areas that relay chemodrive to the respiratory center (168, 169). This likely limits the benefit of drugs that enhance only peripheral chemodrive (136). Most promising are glutamate receptor modulators that may increase neuronal activity in many affected brain stem regions; however, so far, only AMPA receptor modulators have been investigated (116a, 131). Likely, the effectiveness of respiratory stimulants will always be limited at very high opioid doses, and their use has to be balanced with their central nervous and cardiovascular side effects (31, 52).