Severe Hypoxemia Prevents Spontaneous and Naloxone-induced Breathing Recovery after Fentanyl Overdose in Awake and Sedated Rats

Death by opioid overdose is the direct consequence of a rapid and profound depression in breathing,¹  which, if not very rapidly corrected,²  leads to a terminal hypoxic cardiac arrest.³  The substratum of this acute breathing depression is to be found in the combination of different factors. The first mechanism is the direct consequence of the long-described presence of opioid receptors in the aggregates of neurons involved in breathing generation and located in the pons and the medulla.^4–6  These key respiratory neuronal networks, essential for breathing generation, see their rhythmic activity depressed or even abolished in the presence of an opioid agent.⁵  Second, large doses of opioids produce tetanic muscle contractions involving the chest wall and the abdominal muscles,^7–10  a phenomenon well described with fentanyl.¹⁰  This muscle rigidity is produced by a central mechanism, likely mediated in the locus coeruleus,^11,12  that has been shown to decrease chest wall compliance. Along with these changes in respiratory mechanics, sustained contractions of the expiratory muscles⁷  can oppose inspiratory movements.⁸  Finally, laryngeal muscle contractions can lead to a glottis closure,^13,14  adding an obstructive component to an otherwise central apnea.

All of these effects can develop within a few minutes or even seconds in victims of opioid overdose, potentially leading to a terminal hypoxemia and death by pulseless electrical activity or ventricular arrhythmia. The two hypotheses tested in the current study were that the severity of hypoxemia developing during fentanyl-induced apnea will dictate the ability of breathing to spontaneously resume as well as the response to naloxone. Our rationale is that brain stem hypoxia has long been shown to inhibit the medullary respiratory activity^15–17  leading to a central depression of breathing^15,18  in small and large mammals.^16,17,19  As hypoxemia-induced ventilatory depression is a nonopioid-mediated mechanism of breathing inhibition, we hypothesized that if Pao₂ is not restored or maintained above critical levels, via spontaneous gasping, ventilatory support, or rapid recovery of a regular respiratory rhythm, apnea-induced hypoxemia could prevent (1) breathing resumption and (2) effectiveness of naloxone. To test these two hypotheses, we have first investigated the outcome of fentanyl-induced apnea in a nonsedated rat model in keeping with the level of hypoxemia. As this nonsedated model did not allow for the monitoring of arterial blood gases and arterial blood pressure, these studies were completed in spontaneously breathing urethane-anesthetized rats equipped with an arterial line. The combination of these two models has helped us clarify whether hypoxemia-induced ventilatory depression affects the spontaneous recovery as well as the antidotal effects of naloxone after fentanyl overdose.

Fifty male Sprague-Dawley rats (Charles River, USA), weighing 411 ± 20 g (11 to 13 weeks), were used in these studies. The number of rats used for each part of the study, i.e., pilot experiments as well as the unsedated and anesthetized protocols, is given in the protocol and is summarized in Supplemental Digital Content figure 1 (https://links.lww.com/ALN/C201). Animals were randomly allocated to any part of the protocol by a person who was not involved in the conduct of the experiments or the analysis of data. Experiments were always performed in the morning or early afternoon from 10:00 am to 1:00 pm, studying typically two to four rats per session for the unsedated model and one animal only for the anesthetized model (see Protocol section). All experiments were conducted in accordance with the Guide for the Care and Use of Laboratory Animals, 8th Edition (National Research Council [US] Institute for Laboratory Animal Research). The Pennsylvania State University Hershey Medical Center Institutional Animal Care and Use Committee approved the study. The rats were housed at the animal resource services at the Pennsylvania State University College of Medicine (Hershey, Pennsylvania), which conforms to the requirements of the U.S. Department of Agriculture and the U.S. Department of Health, Education and Welfare (Washington, D.C.). The Pennsylvania State University College of Medicine is accredited by the American Association for Assessment and Accreditation of Laboratory Animal Care International (Frederick, Maryland), Animal Welfare Assurance Number A3045-01. Rats were provided with food and water ad libitum, on a standard 12-h (7:00 am to 7:00 pm) light/dark schedule, under the direct supervision of veterinarians.

Fentanyl (Fentanyl Citrate Injection, 1,000 mcg Fentanyl/20 ml [50 mcg · ml^-1], Hospira, Inc., USA) and naloxone (Naloxone Hydrochloride Injection, USP 4 mg/10 ml [0.4 mg · ml^-1], Somerset Therapeutics, LLC, USA) were used in this study; dilutions, when needed, were always done with sterile saline. Two different animal models were used.

The day of study, rats were sedated (3.5% isoflurane) for 15 min to place a heparinized custom-designed double lumen venous catheter in the dorsal vein of the tail. At least 2 h were given for full recovery before doing the study. A custom-designed leak-proof acrylic cylinder (internal volume 1.4 l) was used for determination of ventilation and pulmonary gas exchange at various fractional inspired oxygen tension (Fio₂). Gas was delivered through the inlet port of the plethysmographic chamber using a precision rotameter with a flow of approximately 2.5 l · min^-1.²⁰  The outlet port was connected through noncompliant tubing and to a Fleisch 01 pneumotachograph connected to a pressure transducer (Sensym, DCLX O1DN; Honeywell, USA) for the determination of the flow of air through the chamber. The fractions of carbon dioxide and oxygen were determined (model No. 17630 infrared, and No. 17620 fuel cell analyzers, respectively; Vacumed, USA) in the gas entering and leaving the chamber for computation of carbon dioxide production() and oxygen consumption, as previously described.²⁰  Temperature in the chamber was also continuously monitored via a fast responding thermocouple (Thermalert TH5; Physiotemp, USA). Air could be mixed with nitrogen or replaced by oxygen to create any desired level of Fio₂. In addition, the venous catheter was connected through two extension lines to leakproof adapters, allowing the infusion lines to be connected outside the box. The solution of fentanyl (50 µg · ml^-1) was infused intravenously using a high precision infusion pump (Fusion 100; Chemyx Inc, USA).

All of the analog output signals, i.e., gas flow, percent carbon dioxide, percent oxygen, and temperature were fed into a digital data acquisition system (PowerLab/chart system, AD Instruments, USA). Signals were sampled at 400 Hz, displayed online, and then stored for additional analysis. To calculate respiratory variables from the raw flow trace, a high-pass filter (greater than 0.5 Hz) was used to subtract the direct current voltage component. The “filtered” signal was used for the determination of breathing frequency and to estimate minute ventilation () as previously described.²⁰  To obtain an estimate of from the filtered flow signal, the positive deflections in the plethysmographic trace were integrated over 5-s intervals. The result of this integration was then temperature-corrected based upon the difference between ambient temperatures in the chamber as determined continuously throughout the experiments. Due to the complexity of factors involved in quantitative interpretation of an open-flow plethysmographic signal,^21–23  should be seen as a semiquantitative index represented in the same units as are appropriate for direct measurements of ventilation.

Animals were briefly anesthetized by inhalation of 3 to 5% isoflurane for a few minutes before an intraperitoneal injection of urethane (1.8 g/kg) as previously described. Rats were then tracheostomized, and the tracheostomy was connected to a small dead space two-way valve. Inspiratory flow was measured using a pneumotachograph (Series 1100, Hans Rudolph, Inc., USA). A 7-ml mixing chamber was connected to the expiratory port of the valve, where mixed expiratory gas composition was continuously analyzed (Gemini, CWE Inc., USA). A catheter (PE-50; Fischer, USA) was placed into the right carotid artery to continuously monitor arterial blood pressure and for sampling arterial blood. The arterial catheter was connected to a pressure transducer (MLT844 Pressure Transducer, AD Instruments, USA). Two similar catheters were placed into the left and the right jugular veins for injecting fentanyl and naloxone, respectively.

Body temperature was monitored with a rectal probe (Thermalert TH-5, Physitemp, USA) and maintained around 37°C using a heating pad and lamp. At the end of the experiment, rats were euthanized by a lethal injection of high-dose barbiturate IV (Euthasol; Virbac, USA) into the right heart through the jugular catheter.

Arterial blood pressure, mixed expired oxygen and carbon dioxide fractions, and respiratory flow signals were digitized (PowerLab/chart system) for the computation of minute ventilation, oxygen consumption, and . Blood gases and lactate levels were measured using an i-STAT1 blood gas analyzer (ABAXIS, USA). All signals were visualized online. Data were stored for further analysis using LabChart7.

Preliminary data were collected in six rats using 50, 100, and 300 µg · kg^-1 of fentanyl over 1 min to determine the maximal dose of fentanyl that would produce an immediate, spontaneously reversible apnea. The selected dose was applied to 17 rats. Nine rats were exposed to air, while eight rats were exposed to a hypoxic mixture ranging from 7.3 to 11.3%. Infusion of fentanyl was maintained for 1 min. The time to apnea and the duration of apnea were determined. For all the animals that recovered a rhythmic activity, naloxone (2 mg · kg^-1) was administered at 10 min. Whenever an animal was unable to recover a rhythmic activity, naloxone was administered within 1 min after the median recovery time of air-breathing animals.

Since sedation can alter the response to fentanyl, we first established the dose and rate of fentanyl able to produce an apnea within 10 min in urethane-sedated rats. Our objective was (1) to produce a model that would lead to an apnea after a period of hypoventilation, allowing a low level of Pao₂ to develop at the time of apnea; and (2) to evaluate in this model the effect of hyperoxia.

Seventeen rats were studied in this protocol. Based on the pilot experiments in seven rats (see Results section), the infusion of fentanyl at the rate 2.5 µg · kg^-1 · min^-1 was maintained until minute ventilation would decrease by half to evaluate the effects of naloxone in nonapneic animals. In the other 10 rats, fentanyl was maintained at the same rate (2.5 µg · kg^-1 · min^-1) until an apnea occurred. The inspiratory port of the tracheal valve was connected to a bag containing either air (five rats) or 100% O₂ (five rats); hyperoxia was administered before—and maintained throughout—fentanyl exposure. Arterial blood gases were sampled before fentanyl administration, then within 15 s into the apneic period or during the phase of hypoventilation (for low rate infusion), and just before naloxone administration. Also, arterial blood pressure, which was continuously monitored, was determined at the time of naloxone administration. Whenever naloxone was able to restore eupneic breathing, high-dose fentanyl infusion (25 to 50 µg · kg^-1 · min^-1, i.e., 10 to 20 times the dose used to produce a depressive effect) was infused again 15 min after naloxone as a proof of its efficacy.

Four rats were equipped with an additional arterial catheter placed in one of the femoral arteries, allowing blood sampling without interrupting the recording of mean arterial blood pressure. The left femoral artery and right carotid artery were also exposed and isolated from the vein, nerve, and surrounding tissues, and a transonic flow probe (MA1PRB, Transonic Systems Inc., USA) was placed around both vessels. The transonic flow probes were connected to a perivascular flowmeter (TS420, Transonic Systems Inc., USA). The inspiratory port of the valve was connected to a bag containing different concentrations of a mixture of air and nitrogen. The animals were exposed to this anoxic mixture until breathing stopped, and an arterial blood gas was sampled within 15 s into apnea. Animals were then mechanically ventilated to prevent a cardiac arrest and to allow breathing to resume. Negative inflections in the tracheal pressure signals were considered as spontaneous breaths.

As our data are not expected to be normally distributed, they are presented as median and range, and nonparametric tests were used for statistical analyses. These analyses were conducted using GraphPad Prism 5 (Graphpad Software, USA). Statistical power calculation was conducted before the study. In the unsedated model, anticipating that in air conditions all animals will be responding to naloxone and will display a spontaneous recovery of rhythmic breathing, if 50 to 60% of the animals exposed to a hypoxic mixture are able to recover a spontaneous rhythm or respond to naloxone, a sample of 8 to 11 per group would be required, with a power of 90% and a significance level of 5%. Similarly, assuming that all the sedated animals presenting an apnea after a period of hypoventilation will not respond to naloxone and that hyperoxia may have a rescuing effect in 70 to 80% of animals, four to six animals in each group would be required to identify a statistical difference.

In unsedated animals, our primary outcomes were the onset and duration of apnea (comparison by Mann–Whitney U test). The percentage of animals recovering a regular rhythmic pattern spontaneously and after naloxone was compared between hypoxemic and air-breathing animals using a Fisher’s exact test. In addition, minute ventilation () and were compared at baseline, 10 min after fentanyl infusion (before naloxone injection), and 1 min after naloxone administration within each group using a Friedman test. A Wilcoxon matched-pairs signed rank test was used to compare any of these specific time points, if needed. Comparison between groups was done using a Kruskal–Wallis test for multiple comparison and a Mann Whitney U test for two group comparisons. In sedated animals, the percentage of animals recovering a regular rhythmic pattern spontaneously and after naloxone between the control group and the hyperoxic group was our primary outcome (Fisher’s exact test). In addition, the ventilatory parameters, blood gas values, and arterial blood pressure were compared using a Friedman and Wilcoxon matched-pairs signed rank test for comparison within each group; a Mann Whitney U test was used for comparisons between groups. P < 0.05 was regarded as significant for any of these comparisons that were performed as two-tailed tests.

The selection of the dose of fentanyl was done in a series of pilot experiments (six rats) using a dose of fentanyl of 50, 100, or 300 µg · kg^-1 administered over 1 min—assuming that a dose of 80 µg · kg^-1 could produce a lethal apnea.²⁴  We found that 50 µg · kg^-1 fentanyl infusion produced a rapid apnea that was followed by a spontaneous recovery of a slow rhythmic pattern. A similar apneic response was produced at the dose of 100 and 300 µg · kg^-1 that was always followed by a spontaneous recovery, as described below in more detail for the highest dose of fentanyl infusion (300 µg · kg^-1) that was chosen for our study.

The response to fentanyl infusion (300 µg · kg^-1 infused in 1 min) in air was analyzed in nine rats. An example is displayed in figure 1. All of the animals presented a cessation of breathing at 14 s (median, ranging from 12 to 29 s) after the onset of fentanyl infusion (fig. 2) preceded by an abrupt coma. This apnea was associated with the development of generalized muscle rigidity, which affected the axial muscles (tail, abdominal, chest, back, and neck muscles) followed by a tonic extension of the hindlimbs along with jerking and clear expiratory movements. Every animal spontaneously recovered from this central apnea (median 85 s, ranging from 33 to 141 s) after fentanyl injection (fig. 2). This recovery consisted of a slow reappearance of breaths with reduced tidal volume (no gasping pattern was noted), initially irregular but rapidly becoming regular (fig. 1). As displayed in figure 2, where all individual data along with their median are shown, the level of ventilation emerging after the apnea was very low (median 78 vs. 314 ml · min^-1 for baseline; P = 0.004, Wilcoxon matched-pairs signed rank test) and remained reduced (one third of the baseline value) for the 10-min period during which breathing was recorded. This hypoventilation was associated with a significant decrease in pulmonary gas exchange () from 16.1 ml · min^-1 (median at baseline) to 8.9 ml · min^-1 (10 min into recovery, P = 0.004, Wilcoxon matched-pairs signed rank test). The depression in breathing was, however, out of proportion to the change in metabolism, with a ratio that was significantly decreased (fig. 2). Administration of naloxone (2 mg · kg^-1) increased ventilation immediately (fig. 1) to a median of 513 ml · min^-1 (P = 0.004 when compared to preinjection value, fig. 2). All animals recovered with no clinical neurologic deficits.

Eight rats were studied in hypoxia (Fio₂ ranging from 7.4 to 11.3%). As illustrated in figures 1 and 2, breathing a hypoxic mixture led to an increase in median baseline minute ventilation to 456 ml · min^-1 (P = 0.004, Mann Whitney U test) when compared to air breathing. Yet fentanyl injection still produced a central apnea with the same delay (median 15 s, ranging from 11 to 19 s) as in the air-breathing animals (fig. 2, Mann–Whitney test P = 0.743). However, the recovery from apnea was affected as none of the hypoxic rats, except the one rat exposed to 11.3% O₂ (figs. 1 and 2), recovered a spontaneous rhythmic eupneic activity, nor did they respond to naloxone (fig. 2). Overall, only the rat that was exposed to a hypoxic mixture above 10% (one out of eight rats) recovered and was responsive to naloxone vs. all (nine out of nine) in air breathing (Fisher’s exact test, P < 0.001). Of note, in some animals, a pulse pressure signal was still observed during and after naloxone injection on the respiratory flow signal. However, since the pulse pressure signal superimposed on the flow signal was inconsistently observed even in air-breathing animals, the circulatory status at the time of naloxone injection could not be determined in this model. Blood gases and arterial blood pressure were therefore measured in a sedated model, as presented in the following paragraphs.

As mentioned in the Methods section, the objective of this protocol was to allow a period of hypoventilation to precede the apnea and thus hypoxemia to be present at the time the apnea occurs. In a series of pilot experiments, doses ranging from 2.5 to 50 µg · kg^-1 · min^-1 were infused in six rats to determine the rate of fentanyl infusion that would produce a complete central apnea within 10 min after a period of hypoventilation. We found that infusing fentanyl at a rate of 2.5 µg · kg^-1 · min^-1 for 3 to 5 min was able to depress breathing by half and would stop breathing when infused at the same rate for 10 min. These doses and rates were selected for our protocol to produce either a reversible hypoventilation or an apnea after a period of hypoventilation. As soon as an apnea occurred, fentanyl infusion was stopped. Also, three doses of naloxone, i.e., 0.4, 1, and 2 mg · kg^-1, were investigated. Our aim was to use a dose of naloxone that will be able to prevent any depression in minute ventilation after a dose of fentanyl that was 10 to 20 times the dose producing an apnea. Naloxone at the dose of 2 mg · kg^-1 prevented any new ventilatory depression even when very high doses of fentanyl infusion were used.

Seven rats received an infusion of fentanyl at the rate of 2.5 µg · kg^-1 · min^-1, which was maintained until minute ventilation decreased by half (fig. 3). This took two periods of 52 to 511 s (median 184 s) of infusion each separated by 15 min to develop (fig. 4). More specifically, after the first injection, ventilation returned toward baseline, while a second fentanyl infusion produced a more sustained breathing inhibition (figs. 3 and 4). This depression in breathing was associated with a decrease in oxygen uptake (P = 0.003) and (P = 0.001). As displayed in figure 5, statistically significant hypercapnia and hypoxemia developed, but with no systematic changes in mean arterial blood pressure. A bolus injection of naloxone (2 mg · kg^-1) restored ventilation in every instance (figs. 4 and 5). The effects of naloxone were immediate, occurring no more than a few seconds after a bolus administration. This was associated with a recovery of Paco₂ and Pao₂ (fig. 5). Whenever a new infusion of fentanyl was initiated at the rate of 50 µg · kg^-1 · min^-1 (20 times the rate of fentanyl infusion producing a breathing inhibition before naloxone administration), no further depressive effects on breathing could be observed, as illustrated in figure 3.

Five rats were studied while breathing air during and after an infusion of fentanyl (2.5 µg · kg^-1 · min^-1), which was maintained until an apnea occurred (fig. 4). At this rate, a complete cessation of breathing movements was produced at 9.2 min (time to apnea ranged from 3.4 to 15.1 min). Of note, in two out of five rats, a typical gasping pattern was observed before the animals went apneic. As displayed in figure 4, the cessation of breathing was associated with a statistically significant reduction in mean arterial blood pressure. The most notable changes during the apnea consisted of a reduction in Pao₂ from 79 (range 70 to 95) to 11 (range 6 to 16) mmHg. Naloxone was administered 45 s into apnea, but was unable to rescue any of the animals. Of note, naloxone was administered while a clear pulse pressure was still present on the arterial blood pressure signal (fig. 4). Apnea persisted and led to a cardiac arrest in every animal.

Exposing five animals to 100% O₂ during fentanyl infusion (figs. 4 and 5) produced a progressive depression in breathing leading to an apnea that occurred at a time (median 13.1 min, ranging from 4.1 to 16.1 min) that was not different from air breathing. However, all the animals maintained their Pao₂ well above 100 mmHg at the time of apnea with no difference in Paco₂ between air and hyperoxia (fig. 5). The animals were all rescued within a few seconds after naloxone administration (100% recovery vs. 0% in air breathing, P = 0.008, Fisher’s exact test, two-tailed). Also, arterial blood pressure was maintained around baseline levels at the moment of apnea (fig. 4).

In five rats, breathing nitrogen for 15 s led to an increase in minute ventilation, which rapidly decreased before abruptly stopping, after the generation of several large breaths, as shown in figure 6. Pao₂ determined at the onset of apnea revealed that Pao₂ reached 10 (8 to 12) mmHg at the time of apnea. No spontaneous recovery of breathing occurred. Only when mechanical ventilation was initiated did breathing resume; spontaneous breaths were generated within 10 to 60 s of ventilatory support. Spontaneous ventilation remained sustained thereafter.

We found that, in unsedated animals, the level of hypoxemia reached during fentanyl-induced apnea dictates the ability of rats to “autoresuscitate” and to respond to naloxone. In addition, fentanyl-induced apnea in urethane-anesthetized animals was not associated with spontaneous recovery when Pao₂ decreased below approximately 16 mmHg during apnea. We could also confirm with this model that the resistance to naloxone is a direct consequence of the effects of hypoxemia (fig. 4).

The correspondence between the doses of fentanyl used in rats and those responsible for an opioid overdose in an adult human is difficult to establish. According to the recommendations for conversion of doses (from mg · kg^-1 to mg · m^-2),²⁵  a five times larger coefficient factor should be used to convert doses from an adult human to a 400-g rodent. Accordingly, 300 µg · kg^-1 in an adult rat should correspond to a dose of 60 µg · kg^-1 in a human, while 2.5 µg · kg^-1 · min^-1 for 10 min required to lead to apnea in the sedated model would correspond to a total dose of 0.5 µg · kg^-1 · min^-1. Similarly, the dose of naloxone that was used in our study (2 mg · kg^-1) corresponds to a dose several-fold higher than the maximal doses recommended in adult humans²  even after correcting for the difference in body size. It is the view of the authors that the doses of fentanyl used in this study correspond to doses that would lead to a potentially lethal overdose in humans. These doses were also higher than those used by Ren et al.,²⁴  who found that 80 µg · kg^-1 infused over 1 min was lethal in the group of five rats that he studied. We do not have a clear explanation of why 80 µg · kg^-1 would always cause a fatal apnea, while in our hands, fentanyl at the dose of 300 µg · kg^-1 was unable to produce a fatal apnea in most animals. This dose of 300 µg · kg^-1 was well above the dose required to produce an apnea as established in our pilot experiments. Whether the presence of abdominal contractions during the apnea (fig. 1) when doses of 300 µg · kg^-1 could have mobilized large lung volumes below the functional residual capacity, could have limited the severity of the hypoxemia during the period of apnea, when compared to the lower dosage,²⁴  remains to be investigated.

Since the unsedated model was not equipped with an arterial line, the determination of the circulatory status or the severity of hypoxemia at the time of naloxone injection was not possible to establish. The use of the anesthetized model showed that an effective circulation was still present in the hypoxic animals that did not respond to naloxone.

Important differences exist between our unsedated and anesthetized animal models that must be acknowledged.²⁶  First, in keeping with the difference of doses that depressed breathing in the two models, urethane-anesthetized rats appear to be much more vulnerable to the depressing effects of fentanyl than the awake model. The second difference was the lack of visible muscle rigidity in the urethane-anesthetized rats, while major tonic muscle contractions/contractures involving the back and tail muscles, as well as the abdominal and chest wall muscles, were observed in every awake animal. Similar observations were made by Lui et al.⁹  in rats anesthetized with thiopental, in which opioid-induced rigidity was absent. In contrast, ketamine anesthesia seemed to preserve fentanyl-induced muscle contractures.⁹  The study of the role of muscle rigidity in the ventilatory response to fentanyl is beyond the scope of the current investigation, but chest wall and abdominal muscle rigidity has been suggested to contribute to the severity of the breathing depression produced by fentanyl.⁸  Of note, the dramatic decrease in the ratio, after fentanyl intoxication in both models, reflected the development of a primary depression in breathing occurring independently of the reduction in metabolism typically produced by hypoxemia in small-sized mammals.^22,27 

Hypoxia-induced apnea relies on a complex set of mechanisms, which have long been investigated.^{15,18,28–30}  Indeed, the stimulation of breathing through the arterial chemoreflex produced by hypoxemia^31–34  is counteracted by the depression of the central pattern generator of breathing when a profound reduction in Pao₂ is present, as illustrated in figure 6. Although breathing inhibition by hypoxia has been extensively studied in the fetal and neonatal period,³⁵  this effect remains potent in adults during anoxic as well as ischemic insults.^16,19  In adult sedated sheep,¹⁹  we found that hypoxia-induced apnea was produced at Pao₂ averaging 15 ± 1 mmHg, while values of 13 ± 2 mmHg were observed at the onset of the first gasp. Similarly, in adult awake and chloralose-anesthetized dogs,¹⁶  ventilatory depression develops as soon as Pao₂ reaches 19 mmHg with no influence of Paco₂. These studies show that in adult large mammals, hypoxic ventilatory depression is produced at Pao₂ values that were in ranges similar to those found in our current study. While fentanyl-suppressed hypercapnia induced breathing stimulation,³⁶  hypoxia when reaching values less than 16 mmHg significantly added its depressive effects to those produced by fentanyl, but through a nonopioid mechanism of ventilatory inhibition. Indeed, the on-off depressive effects of hypoxia were very rapid (fig. 6). It has been shown that this depressive effect could be mediated via changes in ion channel activity–induced neuronal hyperpolarization³⁰  and modification in neurotransmitter release³⁵  leading to a disturbance of synaptic interaction within the respiratory neuronal network. These effects are not reversed by naloxone.²⁸  Yet some of these effects are similar to those produced by opioids,^37–39  at least in their mechanisms of transduction, such as an increase in potassium permeability. This mechanism of action has long been supported by the “antidotal effects” of blockers of potassium channels, like doxapram, or phosphodiesterase-4 inhibitors, like caffeine or aminophylline, that increase cyclic adenosine monophosphate against opioid induced respiratory depression (see review⁴⁰ ). Interestingly, some of these agents have also been suggested to oppose hypoxia-induced ventilatory depression.^{18,28–30,41–43}  Regardless of the mechanisms involved, our current data support the view that hypoxia is responsible for a resistance to naloxone after fentanyl-induced apnea when low Pao₂ is present.

The mechanisms leading to breathing recovery at such a high dose of fentanyl are not clear. It was prevented when critical low levels of Pao₂ were reached. Whether the slow frequency and low tidal volume respiration observed after an apnea produced by fentanyl are due to a rapid desensitization of opioid receptors in respiratory neurons previously inhibited by fentanyl or a reconfiguration of the central pattern generators using new neuronal circuits “composed” of cells resistant to opioids¹  certainly remains an outstanding question.

Severe hypoxemia alone produced a decrease in arterial blood pressure associated with a decrease of carotid as well as femoral blood flow in our urethane-sedated rats, as previously reported.⁴⁴  Similarly, when fentanyl was infused with a “slow” rate allowing hypoventilation to develop over several minutes before an apnea occurred, arterial blood pressure fell. Yet at the time of naloxone injection, a significant pulse pressure was still present, while mean arterial blood pressure was reduced by only approximately 28 mmHg (fig. 4). A reduction in medullary blood flow can stimulate brainstem neurons through a local acidosis. A more severe acute reduction in brainstem blood supply, like during a cardiac arrest, for instance,¹⁹  opposes such a stimulation via local brainstem ischemia/anoxia. In the context of a fentanyl overdose, any local hypoxia/ischemia resulting from brainstem hypoperfusion⁴⁴  could have amplified the depression in neuronal activity produced by a reduction in Pao₂. Although we cannot rule out that, in some unsedated animals, a hypoxic depression in cardiac contractility or even a rapidly developing pulseless electrical activity could have prevented the effects of naloxone, the observation that an effective circulation (fig. 4) was still present when naloxone was administered, in the sedated model, suggests that the lack of response to naloxone cannot be explained by its inability to reach the medulla and/or that animals were already in cardiac arrest. Whether high doses of naloxone, higher than 2 mg · kg^-1, could have resulted in a different outcome was not investigated.

Programs of naloxone distribution aimed at providing naloxone kits to the general population have been developed, since a rapid administration of naloxone (intranasal or intravenous) in a prehospital setting is the only effective treatment able to rescue victims found unconscious and in respiratory failure.^45,46  Our current findings support the view that if no ventilatory support²  is provided before naloxone administration, this antidote may be ineffective (or less effective). As there is currently no specific antagonist/antidote that could either promote gasping or counteract the acute effects of hypoxia-induced neuronal depression, lay bystanders, who have free access to naloxone, should always provide ventilatory support before administering naloxone²  in victims of opioid overdose^45,46  with life-threatening breathing depression.

In conclusion, the outcome of fentanyl overdose–induced apnea is dictated by the capacity of a slow but effective spontaneous pattern of breathing to emerge. This new pattern develops in less than 2 min. If Pao₂ reached critically low levels, this rescuing pattern of breathing cannot be produced and is replaced by few gasps that appear unable to produce an autoresuscitation. The response to naloxone was also inhibited, and the outcome was always lethal.

Start-up funds were provided by the Department of Medicine, College of Medicine, Pennsylvania State University, Hershey, Pennsylvania.

The authors declare no competing interests.