Review of CO₂ as a Euthanasia Agent for Laboratory Rats and Mice

Abstract

Selecting an appropriate, effective euthanasia agent is controversial. Several recent publications provide clarity on the use of CO₂ in laboratory rats and mice. This review examines previous studies on CO₂ euthanasia and presents the current body of knowledge on the subject. Potential areas for further investigation and recommendations are provided.

An ideal euthanasia method meets several criteria both for the animal as well as for the agent used, as previously described.¹¹ Ideally, euthanasia involves the capture and restraint of animal with minimum distress and the induction of immediate and permanent insensibility with no, or minimal, distress to the animal. The ideal method is aesthetically acceptable to the public and to the person administering the procedure, cost-effective, and easily administered by nonveterinary personnel without extensive training. Finally, this ideal agent would create no, or minimal, risks to the person administering the procedure, cause no damage or residues that might interfere with examination or future use of the carcasses, and generate no residues that might create a hazard to other animals following the final disposal of the carcasses.¹¹

Several euthanasia techniques for laboratory rats and mice are either acceptable or acceptable with conditions as described in the American Veterinary Medical Association (AVMA) guidelines.⁶³ These methods are often categorized by methodology: injectable, inhalant, or physical methods. Classification of agents as acceptable compared with acceptable with conditions or not acceptable was based on the evaluation of 14 different criteria for each method of euthanasia. One of the most commonly used methods of rodent euthanasia is inhalation of CO₂ because it reliably and rapidly induces loss of consciousness with minimal safety concerns for personnel, including the low potential for human abuse. CO₂ euthanasia is a relatively simple method to use and does not require highly specialized equipment or considerable training for personnel to perform successfully. There are no drug residues that could affect predators or scavengers should the carcass be consumed (allowing rodents to be ‘repurposed’ as food for carnivores),¹⁴ and there are no significant safety concerns with using CO₂ in the laboratory setting.³ However, the use of CO₂ as an agent for euthanasia has come under intense scrutiny over the past few decades.¹⁰⁵ Evidence of aversion to CO₂ gas, clinical reports of pain at high concentrations, and gasping due to air hunger experienced during exposure are reasons why CO₂ is not currently considered an optimal euthanasia agent. In 2006, and again in 2013, a group of experts gathered in Newcastle, United Kingdom, to discuss the use of CO₂ as an agent for rodent euthanasia.³⁷ The current review analyzes the 2006 Newcastle report in light of new information, provides recommendations based on the findings from these studies, and outlines areas requiring further investigation. The purpose of the review is to provide further guidance to the scientific community and ethical review groups regarding the use of CO₂ as a euthanasia agent.

The Newcastle Reports

The 2006 Newcastle report represents a consensus opinion of 12 people representing research in CO₂ euthanasia and experts in laboratory animal welfare and animal care.³⁷ They raised 8 issues concerning the use of CO₂ for rodent euthanasia, with salient conclusions divided into 4 categories. The first category addressed problems perceived in using CO₂ for euthanasia, including the lack of an ideal dose of CO₂, given that exposing animals to concentrations of over 50% leads to mucosal pain prior to loss of consciousness, and the expression of aversion and dyspnea with slow chamber-fill rates.³⁷ Second, the authors of the 2006 Newcastle report concluded that, when CO₂ is used, the avoidance or minimization of pain and distress is more important than speed of euthanasia. Therefore, a slow chamber-replacement rate, which invokes less pain response, is recommended, and the addition of O₂ should be considered because it may reduce the welfare problems associated with dyspnea.³⁷ Third, the authors concluded that alternative gases may be useful, but information on agents such as argon, nitrogen and volatile anesthetic agents was insufficient to advocate for these agents as acceptable replacements for CO₂ euthanasia.³⁷ Last, the authors of the 2006 Newcastle report concluded that additional research was required to address the many questions remaining regarding CO₂ use for euthanasia of mice.³⁷

In 2013, a group convened to update the Newcastle Report.³⁸ This group discussed multiple euthanasia techniques including updating information on CO₂ euthanasia. The resulting recommendations for CO₂ use were essentially the same as in the previous report, except that the use of O₂ with CO₂ was now discouraged, given that combining the gases does not consistently improve animal wellbeing and can increase the time to unconsciousness and death.³⁸ The participants in the meeting acknowledged that all currently available euthanasia techniques will inevitably involve some degree of pain or distress or both, and they concluded that CO₂ cannot be considered a humane method of euthanasia for rodents and that developing replacements is an essential goal.³⁸ Interestingly, despite the concern regarding the use of CO₂ euthanasia, a survey of the conference attendees revealed that more than 35% used CO₂ as a euthanasia agent for mice and rats, and CO₂ was the most frequently used euthanasia technique.³⁸

Stress and Pain Assessment

When selecting a euthanasia agent, it is imperative that a humane method is chosen that induces a rapid, painless, and distress-free death.⁶³ Differentiating stress from distress is difficult. Although the effects of stress may be positive, negative, or inconsequential, distress threatens the wellbeing of the animal and is inherently detrimental.²⁰ Stress can be defined as a “real or perceived perturbation to physiologic homeostasis or psychologic wellbeing.”²⁰ Typically, stress is a response to a brief incident and leads to no long-term alteration in physiologic or psychologic changes. In contrast, distress is associated with prolonged exposure to a stressor and leads to altered physiologic or psychologic states.²⁰ The International Association for the Study of Pain has defined pain in humans as “an unpleasant sensory and emotional experience associated with potential or actual tissue damage.”²⁰ In the current review, we have not changed any cited manuscripts’ use of the terms stress, distress, or pain, but readers should interpret the statements in light of these definitions.

A critical concept to consider is that stress, distress, and pain occur while an animal is conscious. If euthanasia is not performed by using a physical method (such as decapitation or cervical dislocation), it is generally performed by using an overexposure to an anesthetic agent. For this reason, it is essential that scientists who are evaluating the wellbeing of animals during the euthanasia process ensure that they are familiar with the different stages of anesthesia and how those stages correlate to conscious and unconscious responses to noxious and distressing stimuli. There are 4 well-recognized stages of anesthesia.^41,104 Stage I is defined as the period of time from induction to loss of consciousness. Stage II is the period of time from loss of consciousness to loss of response to stimuli and is referred to as the excitatory phase, because of the reflexive muscle responses that occur during this period of time. Human studies have found that amnesia and unconsciousness representative of stage II anesthesia occurs at half of the anesthetic dose that it takes to eliminate movement, consistent with the progression of the patient through the accepted stages of anesthesia.⁵ In Stage III of anesthesia, the person or animal is at a surgical plane of anesthesia, with amnesia, analgesia, and muscle relaxation achieved. Stage IV of anesthesia is generally avoided during surgical interventions, because it involves cessation of cardiovascular responses and respiratory function, but it is the desired state when euthanasia is the end goal. From the perspective of the animal, there is no difference in anesthesia induction for a surgical procedure or anesthesia induction for euthanasia. For the animal, only the physiologic changes and behavioral responses experienced during stage I are of concern. The interpretation of movement or response to stimulation as a conscious activity must be carefully assessed, as these behaviors often involve spinally mediated reflex activity, which can occur after a state of unconsciousness is achieved during stage II of anesthesia. But, likewise, dismissal of gross behavior, such as rearing and activity, through misapplication of the term “excitatory phase associated with the induction of anesthesia”⁶⁷ is inappropriate when the behaviors occur during stage I of anesthesia and the animal, therefore, is still fully conscious. Careful understanding and definition of the stages of anesthesia will allow appropriate study design to generate results that will facilitate the ongoing discussion regarding animal wellbeing during anesthesia or euthanasia, regardless of purpose.

Unfortunately, defining when an animal becomes unconscious (that is, moves from stage I to stage II of anesthesia) is not a simple task. Unconsciousness in animals has historically been defined as a loss of righting reflex (LORR), also known as loss of position (LOP).⁷⁵ However, this definition is not based on sound scientific evidence that LORR/LOP is truly the demarcation point for loss of consciousness, given that the loss of reflex responses varies according to species, type of reflex, and anesthetic agent.⁵ Therefore, awareness of the difference between being nonresponsive to external stimuli and the loss of consciousness is critical for animal studies in which we try to define whether animals are experiencing stress, distress, or pain. If a common complaint about CO₂ euthanasia is that the process “looks like the animal is in stress and/or pain,” is the observational period truly representative of the time in which the animal is conscious? Movement, vocalizations, and gasping that occurs after LORR or loss of consciousness may not be indications that the animal is experiencing stress or pain.

CO₂ causes pain at high concentrations.

CO₂ was used in human medicine for the treatment of psychosis as early as the 1920s.⁶⁶ Early observations showed that patients exposed to a 30% CO₂ and 70% O₂ mix required about 35 to 45 respirations to induce narcosis.⁶⁹ During this time, the patients hyperventilated, and the CO₂ exposure was considered to be unpleasant, causing stress, electrocardiogram alterations, muscular tremors, sweating, and gasping in some patients.⁶⁹ More recent work examined a single breath of 4 concentrations of CO₂: 5%, 25%, 35%, and 50%.⁵⁰ Subjective and somatic symptoms of anxiety increased in a dose-dependent manner, with ACTH and noradrenaline responses beginning at 35% CO₂.⁵⁰ Subjects were unable to take a full inspired vital capacity breath of 50% CO₂.⁵⁰ Evidence to support that high concentration of CO₂ is painful is also derived in part from human studies on nasal CO₂ exposure.^4,27,108 Contact of CO₂ with the nasal mucosa leads to the production of intracellular carbonic acid, which decreases pH and thus causes pain.⁴ One study showed self-reported increasing pain intensity after single breaths of increasing CO₂ concentrations of 50% to 100% (lower concentrations were not tested).⁴ Another study evaluated 7 s of nasal exposure to 35.5%, 53%, and 70% CO₂ and found that pain intensity was proportional to CO₂ concentration.¹⁰⁸ Interestingly, pain intensity peaked at 3 to 4 s and then rapidly faded.¹⁰⁸ Similar to human findings of pain, studies in rats show increased firing of medullary dorsal horn neurons, an area of the spinal cord known to respond to noxious chemical stimulation, in response to CO₂ inhalation.⁸⁹ The response magnitude of neural firing was increased at 25% CO₂, and a near-linear increase occurred between 37% and 87% CO_2, suggesting that as CO₂ increased, pain increased.⁸⁹ Another study identified acid-sensing ion channel 1a (ASIC1a) as a critical mechanism in the amygdala for detecting alterations in pH that lead to anxiety and fear.¹¹⁰ Mice without this channel do not develop the anxiety associated with CO₂ exposure that is observed in mice with an intact channel.¹¹⁰

Stress hormone response to euthanasia.

Any stressful situation, including pain, induces a cascade of activity known as the stress response. The body responds to acute stress through 2 pathways: the sympathetic–adrenal–medullary system (SAM) and the hypothalamic–pituitary–adrenal axis (HPA). When stress is perceived by the brain, the SAM responds almost immediately by releasing norepinephrine mostly from sympathetic nerve endings and epinephrine from the adrenal medulla. Concurrently, along the HPA, the stress response begins in the amygdala, initiating corticotrophin releasing factor to be secreted by the hypothalamus. This hormone then stimulates the release of ACTH from the pituitary. ACTH acts on the adrenal cortex to release corticosterone. The release of these chemical messengers results in increases in blood pressure and heart rate as part of the sympathetic response. To evaluate this stress response, components of the HPA axis and SAM hormones can be measured. The intensity of the stress response can be measured through HPA axis activity, given the evidence of graduated responses to different levels of stressors (in these references, pain and distress are used as the stressor).^40,51 For example, physiologic levels of ACTH increase 10- to 50-fold in response to different stressors.^1,8,24 The variable ACTH response does not translate to corticosteroid levels, because only marginal increases lead to maximal corticosteroid secretory rates.⁵¹ Therefore, measuring corticosterone to differentiate stressor intensity is sufficiently sensitive for quantifying the effects of mild stressors only, as shown by exposing mice to increasingly novel situations.⁴⁰ In contrast, for highly stressful situations, as seen in rats exposed to foot shock, corticosterone response was an ineffective measurement tool for characterizing the stress response.³¹

Another complication of using corticosterone as a measure of stress is the delayed onset of increases in hormone measurement. In a study involving humans that took a single breath of 35% CO₂, peak levels of cortisol were not reached until 15 min after exposure, whereas maximal levels of ACTH and norepinephrine were achieved within 2 min.⁵⁰ In other studies, ACTH levels in mice increase within approximately 30 s after exposure to a stressor such as CO₂ gas,^36,93,97 and norepinephrine is elevated within 5 s of CO₂ exposure.¹⁵ Rats respond in a similar fashion: it took approximately 4 min for corticosterone levels to increase in response to stressful events.^25,28,96 Given the delayed response of corticosterone, we believe that ACTH, norepinephrine, and epinephrine are better indicators of differences in stress during euthanasia.

Behavioral evidence of stress or pain during euthanasia.

Evidence of stress, distress, or pain during procedures including euthanasia can be observed as a state of behavioral arousal (that is, a ‘fight-or-flight’ response) in the animal. The most common behaviors during euthanasia include flight responses, through which rodents try to escape from the situation.^{5,13,75,77,78,81,82} Behaviors such as running, escape behaviors, rearing, and approach-avoidance preferences can be used to determine whether mice are fleeing from a stressor.^{5,13,75,77,78,81,82} Behavioral assessments also can include changes in activity, such as sniffing, grooming, and vocalization, in response to a stressor.^13,78,81,82 In a study that examined escape behavior, righting reflex, and pedal withdrawal reflex to toe pinch during euthanasia, mice euthanized with CO₂ were insensible when recumbent, whereas mice that were euthanized with isoflurane had a pedal withdrawal response when initially recumbent.⁷⁸ Because movement and response to stimuli can occur without supraspinal structures,⁵ pedal withdrawal may be an inadequate method for assessing consciousness. Another study that compared responses to a prefilled chamber with a fixed flow rate of CO₂ (20% chamber replacement rate [CRR]), using behavioral responses and blood gas analysis demonstrated differences between treatments in the time to ataxia, immobility, loss of pedal reflex, and respiratory arrest.⁴² The authors concluded that these responses were due to the anesthetic effect of CO₂ and that no distress occurred with either method.⁴² In addition, several behavioral studies have analyzed differences in aversion or avoidance to CO₂ compared with halogenated gases; these studies are discussed in the section on halogenated gases.

Neural response to euthanasia.

A third method of assessment examines neural responses during euthanasia. The minimal alveolar concentration (MAC) of an inhalant anesthetic agent is defined as the concentration that prevents purposeful movement in 50% of subjects in response to a supramaximal noxious stimulus.^6,80,86,99 MAC does not necessarily assess hypnotic properties and is more related to anesthetic effects on the spinal cord rather than effects on the brain. However, a derivative of MAC is MAC-awake, the point at which anesthetized patients are able to respond to verbal commands; MAC-awake represents a surrogate for consciousness during recovery from anesthesia. MAC-awake is estimated to range between 34% to 68% of MAC, depending on the anesthetic agent.⁶ Another criticism of MAC is that it relies on a subjective determination of the experimenter.^80,86 The MAC of CO₂ is 403 mm Hg.¹⁶ Further research is required to determine the CO₂ MAC during euthanasia to add perspective to this value.

Another method for evaluating depth of anesthesia is electroencephalography (EEG). EEG—the recording of the voltage between 2 electrodes either attached to the head or implanted surgically¹⁰—has been used for roughly 30 y to assess the effects of intraoperative anesthesia. Anesthetic agents produce noticeable changes in the electrical potentials recorded from the brain or scalp. In human anesthesia, EEG has become the “most widely evaluated neurophysiological tool to assess depth of anesthesia” to prevent the risk of intraoperative awareness not detected by monitoring of anesthetic depth by using blood pressure and heart rate.^53,86 In animals, EEG patterns have been used to assess various anesthetic agents as well as euthanasia by decapitation, cervical dislocation, isoflurane, potassium chloride, or CO₂ overdose.^21,76,86,107 In addition, sensory and cognitive event-related potentials measured by EEG have been used as reliable biomarkers for assessing the effects of various analgesic drugs to treat pain.^33,34,64,90 The normal, awake EEG is characterized by rhythmic α wave activity.⁸⁶ In general, as depth of anesthesia increases, activation of the cortex by the ascending reticular activating system diminishes, indicated by slowing of the EEG toward a high-amplitude, low-frequency, slow-wave pattern.^80,86 Another way to look at EEG output is to examine burst suppression (BS). The awake EEG BS pattern is characterized by intermittent isoelectric periods interspersed with high-voltage bursts usually lasting 1 to 10 s.⁸⁰ BS or intermittent electrical activity interspersed with near-complete depression of cortical electrical activity can be indicative of a nonspecific reduction in cerebral metabolic activity.⁸⁶ A previous study evaluated the effects of halothane, isoflurane, sevoflurane and desflurane in rats by using EEG.⁸⁰ Burst suppression was found in the EEG recorded from all agents except halothane. Therefore, halothane led to significantly less depression of cortical activity with halothane than did the other inhalant agents, demonstrating the ability to use EEG to qualitatively differentiate between anesthetic mechanisms of action.⁸⁰

In addition, several studies have shown that a reduction in α:δ brain-wave ratios coincides with LORR/LOP in chickens.^70,75 For example, as the level of consciousness decreases, α and β waves are suppressed, and δ and θ waves increase.¹⁰ In that study, unconsciousness was associated with high-amplitude, low-frequency activity of δ and theta waves in ducks, turkeys, layer hens, and broilers.¹⁰ In a study that evaluated the physiologic response of laying hens during “whole-house killing,” EEG responses strongly associated with unconsciousness were used to determine time to loss of consciousness.^10,70

Only a few studies have evaluated the EEG response during euthanasia in rodents.^21,58,107 Two studies focused on the EEG response to decapitation in rats and found that cortical activity was present for 15 to 20 s after decapitation.^58,107 However, the electrical activity was similar to that seen in anesthetized animals, so it is unlikely that the pattern represented consciousness.¹⁰⁷ Another study evaluated both EEG and visual evoked potentials (VEP) as measures of cortical function in mice during various methods of euthanasia and demonstrated rapid disruption of cortical function after decapitation, 100% CO₂ through positive pressure ventilation, and cervical location.²¹ Although EEG is a useful tool for capturing brain activity, its utility in determining the humaneness of euthanasia methods is limited, given that EEG and VEP may be unable to provide definitive answers regarding the precise onset of unconsciousness.⁷⁵ For example, evoked potentials measure the response of the nervous system to stimuli, and VEP are used to test the visual pathway from the retina to the occipital cortex, measuring nerve conduction. Although VEP loss is associated with brain death, the visual cortex remains responsive during desflurane anesthesia; therefore VEP loss does not necessarily correlate with the level of consciousness.⁷⁵ Although often descriptive, the use of EEG as an objective measure of unconsciousness, or determination of state of consciousness, especially for decision-making related to the humaneness of euthanasia methods, is an area of research that requires further exploration.

Cardiovascular response to euthanasia.

Pain and distress cause increased levels of circulating catecholamines, with subsequent increases in heart rate and blood pressure.^{47,48,55,84,95} A primary concern with using cardiovascular responses as markers of pain and distress is that it is unclear whether increases directly correlate to pain and distress. In addition, cardiovascular changes can be caused by the anesthesia or due to organismal differences such as sex.^57,103 In addition, CO₂ has direct myocardial action and causes hyperkalemia, thus further disrupting myocyte activity.⁴⁴ Other studies have examined the use of heart rate variability as a marker of pain.⁵⁷ A recent review concluded that no validated objective markers of nociception or pain recommended for clinical use in humans are available currently, although the authors noted that the analysis of heart rate variability potentially could be developed for this purpose.²⁶ Certainly additional studies in this area are required to evaluate the benefits of heart rate variability as a tool for assessment of pain and distress.

Few studies have examined the cardiovascular response of rats or mice during euthanasia.^12,13,100 As expected, these studies found elevations in heart rate and blood pressure during the euthanasia process. The injection of a pentobarbital–phenytoin euthanasia solution led to an immediate increase in both heart rate and blood pressure in response to the handling associated with injection restraint and to injection site pain.¹² Cardiovascular values differed between 2 studies examining CO₂ euthanasia in laboratory rats and mice.^13,100 In these studies, rat cardiovascular values were increased at the beginning of the euthanasia process, whereas in mice, the increase was delayed.¹⁰⁰ Because any form of stress, including handling and transport of an animal, can lead to increased cardiovascular values, the difference in the 2 studies can be explained by the fact one study allowed the mice to acclimate to the euthanasia chamber set-up prior to turning on the CO₂. One study also examined the effect of isoflurane euthanasia of mice on cardiovascular parameters.¹² Similar to CO₂, increases in heart rate and blood pressure in response to isoflurane were delayed.¹² Finally, the heart rate and blood pressure responses to euthanasia by either pentobarbital–phenytoin euthanasia solution, isoflurane, or CO₂ were compared and showed no significant differences between 4 different (15%, 30%, 50%, 100%) CO₂ CRR; differences between pentobarbital–phenytoin euthanasia solution, isoflurane, and CO₂ included an increase in heart rate and a decrease in blood pressure in the mice given the pentobarbital–phenytoin euthanasia solution.¹² The cardiovascular changes in all of these studies demonstrate that stress was associated with all of the procedures, but stress levels did not differ as a result of the technique used.

Examining Optimal CO₂ Chamber Replacement Rates

Recent studies have examined alternative methods to CO₂ use for rodent euthanasia that address 2 important issues raised by the Newcastle reports: CRR and CO₂ concentration.^12,13,77,91 It is important to recognize that because CO₂ is heavier than air, CO₂ will sink to the bottom of the cage during the euthanasia process. In one study, in addition to cardiovascular data showing no significant differences in response to CRR between 15%, 30%, 50%, and 100% CO₂, there were no significant differences in mouse activity, lung histology, or ACTH levels.¹³ The only significant differences were the length of time that the mice were stressed, which was inversely proportional to the CO₂ CRR.¹³ CO₂ levels in the study, measured 1 to 2 cm above the cage floor, did not reach concentrations that are considered painful prior to full recumbency, suggesting mice did not experience pain during the euthanasia procedure.¹³ Another study, while maintaining CO₂ below painful levels (below 40%), attempted to determine optimum CO₂ CRR to minimize the possibility of dyspnea.⁷⁷ Compared with 20% and 30% CRR, a flow rate of 50% CRR, while holding the CO₂ cage concentration just below 40%, minimized the interval between the onset of labored breathing and recumbency. However, even at the 50% flow rate, mice experienced more than 30 s between the onset of dyspnea, as demonstrated by gasping, and the most conservative estimate of insensibility.⁷⁷

One recent study in Sprague–Dawley rats assessed the potential distress associated with prolonged exposure to CO₂.⁴³ Rats in that study were exposed to 20% CO₂ for 5 min and then allowed to recover. Baseline behavioral scores (social interaction and open field) were compared with those after gas exposure. When tested immediately after exposure, rats that had been exposed to 20% CO₂ exhibited decreased exploratory behavior in the open field and were less likely to interact with a strange rat, suggesting distress secondary to the gas exposure.⁴³ This same study also demonstrated a significant increase in the cardiovascular response of rats euthanized with 10% CO₂ CRR as compared with higher concentrations.⁴³

These observations suggest that rodents euthanized with slow CO₂ CRR have a longer period of distress prior to unconsciousness, with no clear evidence that they experience less pain, compared with mice euthanized with fast CO₂ CRR.^13,77 The studies showed that faster flow rates can be used without inducing pain. In contrast to behavioral observations,¹³ a euthanasia study using a HiRoad euthanasia chamber (Lab Etc., Clayton, DE) and comparing slow-flow (30%) CO₂ with fast-flow (70%) CO₂ found an increase in anxious behaviors in CD1 mice with fast-flow CO₂.⁹¹ This finding shows the importance of potential differences between strain, age, and technique used during the euthanasia process, thus potentially affecting the overall conclusions drawn from any individual study.

Use of O₂ with CO₂

Provision of supplemental O₂ during CO₂ euthanasia has been posited to ameliorate dyspneic behaviors during euthanasia. In male Wistar rats exposed to 125% CO₂ CRR with 62.5% O₂ CRR or without O₂,²³ the abnormal high activity, excitation, and agitation phase of the euthanasia process was totally eliminated, and rats did not gasp during the procedure, when O₂ was given with CO₂. In contrast, the rats without O₂ had a statistically significant increase in both of these activities.²³ The conclusion of the study was that the addition of O₂ the prevented the negative effects of CO₂, but rats had a prolonged time to death and may have been conscious longer when given O₂ in addition to CO₂. In contrast to this single study, many studies have shown no significant benefit from the addition of O₂ during CO₂ euthanasia. However, many of the previous studies were confounded by alterations in CO₂ flow rate with the addition of O₂ or did not show decreases in dyspneic behaviors.^2,42,46,54 For example, in a study that examined CO₂ and O₂ mixtures for euthanasia of rats and mice,⁴⁶ animals were exposed to 100% CO₂ or 80% CO₂ with 20% O₂. The presence of O₂ doubled the time to unconsciousness in rats and mice but appeared to decrease urination, defecation, and labored breathing. Whether the different CO₂ replacement rates, different CO₂ concentrations, or the addition of O₂ altered the results is unclear. In addition, increased hemorrhage and edema was associated with O₂ addition to CO₂.^2,27,46 In one study, the hemorrhage was postulated to induce a feeling of drowning, and the addition of 20% chamber volume per minute of O₂ was not recommended as a refinement to CO₂ euthanasia.² Two studies evaluated aversion to CO₂ with and without O₂.^54,62 Aversion to CO₂ was unaffected by the addition of O₂ when no incentive was provided⁶² and mildly reduced aversion when a treat was provided at slow CRR of 14.5% CO₂.⁵⁴ On the basis of these studies, the AVMA Panel on Euthanasia concluded that there appears to be no advantage to combining O₂ with CO₂ for euthanasia.⁶³

Alternative Gases and Halogenated Gases

Argon, nitrogen, nitrous oxide, and isoflurane have been tested either as alternative gas euthanasia methods or in combination with CO₂ to euthanize rodents.^{17,61,68,77,81,83,98,102,109} The mechanism of euthanasia using argon and nitrogen is by displacement of O₂ to lower than 2% in the chamber, leading to anoxia. Argon euthanasia led to an increased elevation in the heart rate of Sprague–Dawley rats, compared with CO₂ euthanasia.¹⁷ In that study, argon was given at 50% CRR and CO₂ at 10% CRR. Sprague–Dawley rats were observed to have back arching, gasping, and hyperreflexia or seizures during argon euthanasia.¹⁷ In addition, another study showed severe aversion to argon.⁸³ Nitrogen at approximately 100% caused hyperreflexia and was slow to produce unconsciousness and death in Holtzman Sprague–Dawley rats.⁹⁸ Concern regarding animals being hypoxic prior to unconsciousness led the AVMA Panel on Euthanasia to conclude neither nitrogen nor argon is an acceptable euthanasia method for rodents.

Nitrous oxide has a MAC of greater than 100%, rendering it impossible for use as a sole agent for euthanasia. Nitrous oxide in combination with isoflurane or CO₂ led to 18% and 10% reduction. respectively, in time to LORR in mice.¹⁰² This effect may be an improvement with either euthanasia agent, given that it may shorten the exposure to potential stress, distress, or pain with these procedures. However, nitrous oxide has a higher human abuse potential than either CO₂ or isoflurane alone. Further research on the potential animal welfare benefits are required prior to conclusions regarding the use of nitrous oxide for euthanasia of rodents.

Because rodent euthanasia using exposure to increasing concentrations of CO₂ was believed to cause pain during the euthanasia process, one recommendation was to replace the use of CO₂ with halogenated gases. To assess whether halogenated gas anesthesia followed by CO₂ is an improvement over CO₂ alone, several studies have examined aversion as a surrogate for the euthanasia processes. In several elegant studies, rodent aversion to different gases was analyzed.^{61,68,77,81,83,109} In one study, Wistar rats were given food rewards and exposed to CO₂; whether they remained in the chamber and ate the treats or left when exposed to CO₂ was measured.⁸³ In that study, and others, independent of the CO₂ CRR, the concentration at which rats and mice leave is similar (about 12% to 15%).^77,81,83 The results of these CO₂ studies are consistent with human studies, which show the presence of dyspnea at 8% CO₂.^29,65 Compared with findings from human studies, CO₂ likely is only causing distress in the rodents at a 15% concentration, because pain likely does not occur until CO₂ reaches 40%.⁷⁷ In another aversion model, when comparing isoflurane with CO₂, Sprague–Dawley rats tested in a light-dark box escaped to a lit compartment 100% of the time when exposed to CO₂ but only 44% of the time when exposed to isoflurane.¹⁰⁹ Interestingly, after repeat exposure to isoflurane, 94% of rats escaped to the light side, indicating that a second isoflurane exposure was more aversive than the first.¹⁰⁹ Rats have an innate CO₂ receptor in the brain,¹⁰¹ but there is no evidence for a similar isoflurane receptor, suggesting that the response is a learned behavior in the Sprague–Dawley rats. Based on the results of these studies and others, the Canadian Council on Animal Care (CCAC) Guidelines on Euthanasia of Animals used in Science and attendees at the 2006 and 2013 Newcastle meetings on euthanasia recommended the use of anesthetics prior to CO₂ euthanasia.^19,37,38 Surrogate studies, such as regarding the use of aversion or avoidance, are beneficial for defining aspects of the response to inhalant agents, but they do not adequately define whether pain or distress is experienced, nor do they replicate the euthanasia experience in its entirety. Aversion does not mean that the animals are in pain, distress, or in a negative affective state; it only represents that the animal does not want to be in the area because they do not like something and that they have the ability to exit the area or otherwise indicate their preference. Therefore, we believe that the conclusions of the prior studies may be premature and warrant further information to translate results regarding aversion to euthanasia.

Further studies have augmented the aversion studies, directly comparing CO₂ euthanasia and halogenated gas euthanasia. Isoflurane, as a euthanasia agent in naïve animals, is not as innocuous as proposed in the aversion studies. For example, a study examining 3 levels of insensibility during isoflurane euthanasia found that all mice had an escape response, purposeful movement, and a pedal withdrawal response.⁷⁸ Interestingly, mice euthanized with CO₂ exhibited these same responses much less frequently.⁷⁸ In addition, response may differ by sex, given that anesthesia with a brief exposure to isoflurane prior to decapitation elevates total plasma corticosterone in female, but not male, Wistar rats.⁹ In addition, male C57BL/6 mice exposed to either isoflurane or CO₂ for euthanasia had equally increased ACTH responses, suggesting that the stress associated with either technique was similar.¹² That study also found no significant differences in behavior or cardiovascular values that would be indicative of an increased stress response with CO₂ when compared with isoflurane.¹² In another study, CD1 mice were anesthetized with isoflurane prior to CO₂ euthanasia and compared with euthanasia with 20% and 100% CO₂ CRR.¹⁰⁶ The authors found isoflurane increased behavioral and neuromolecular signs of stress, compared with CO₂.¹⁰⁶ Finally, another study examining CD1 mice showed significant increase in anxious behaviors in mice either euthanized with isoflurane or exposed to a brightly lit chamber, compared with 30% CO₂ flow-rate euthanasia.⁹¹ Although differences in aversion are present between CO₂ and isoflurane exposure, the CO₂ euthanasia process does not appear to elicit a response that is more stressful than that to isoflurane. In fact, some evidence suggests isoflurane prior to CO₂ is more stressful.¹⁰⁶ Therefore, there appears little justification to recommend the replacement of CO₂ euthanasia with isoflurane on the basis of the level of stress or distress. In addition, the use of halogenated gas presents safety concerns for the personnel conducting the euthanasia. One difference noted is a greater incidence of histologic damage in the lungs associated with CO₂ euthanasia compared with isoflurane.¹² Under circumstances requiring lung histologic analysis, the use of isoflurane is recommended prior to or instead of CO₂.

Embryonic, Fetal, and Neonatal Mice

Scientific or management needs may call for the euthanasia of mice for harvesting of germplasm and embryos, the euthanasia of gravid mice and their fetuses, or the euthanasia of mice before weaning. The literature relevant to these specialized applications of CO₂ is summarized in this section.

Investigators working with mouse germplasm and embryos typically euthanize animals by using cervical dislocation. An alternative method to performing cervical dislocation is the use of CO₂, which similarly does not chemically contaminate the germplasm or embryo. Support for this practice was provided by a study that compared cervical dislocation with CO₂ and did not find differences in embryo yield or viability.⁴⁵ However, similar work revealed differences in sperm and oocyte functionality when animals were euthanized with CO₂.³⁹ Further studies are required to determine the cause of these differences and whether CO₂ can be used to replace cervical dislocation for germplasm and embryo harvest.

In 2 studies of euthanasia of fetal mice in utero, CO₂ was an effective means of euthanasia of pregnant animals and fetuses as measured by cessation of fetal heartbeat.^56,79 Fetal death, measured by using ultrasonography to assess cessation of heartbeat, was prolonged (exceeding 20 min) regardless of the method used, pointing to fetal resilience to CO₂.⁷⁹ The method recommended by the authors of that study, intraplacental injection of sodium pentobarbital, involves removing animals from the uterine environment. Because fetal animals are not conscious^18,72-74 and because separating them from the uterine environment might induce consciousness, separate euthanasia methods for dam and fetus are not advised unless the fetuses are to be manipulated.⁷¹

The euthanasia of neonatal mice has been addressed in 2 recent papers.^56,92 Both agree that the physiologic responses of neonatal mice to low O₂ concentrations and their resistance to the effects of hypercapnea prolong the time to death when CO₂ is used. In one study,⁹² neonatal mice reached apparent insensibility within 1 min, but extended exposures were required to ensure death in 0- to 8-d-old mice. Preweanling animals differed in their susceptibility to CO₂ by genetic background but not by sex.⁹² Outbred preweanling mice were more susceptible to CO₂ euthanasia between days 0 and 8 compared with inbred preweanlings.⁹² In another study,⁵⁶ mice exposed to CO₂ were closely monitored for LORR and cardiac arrest. The LORR after CO₂ exposure was longer than that recorded previously,⁵⁶ but death as an endpoint was not always reached, because some of the pups recovered after cardiac arrest.^56,92 The author of 2 other papers reached similar conclusions, namely, that neonatal mice are resistant to the effects of CO₂.^7,94

For fetal and germplasm viability, labs should use the method that yields the best results in their hands. For some, a physical method might be optimal, whereas for others, CO₂ might be appropriate. Gravid animals can be euthanized as though they were nongravid; fetuses die when the dam does and are never conscious unless separated from the dam. When CO₂ is used to euthanize mice younger than 14 d, prolonged exposure times are necessary, and the application of a secondary euthanasia method, such as decapitation, is advised.

Future Directions and Gaps in Our Understanding

Several areas regarding the use of CO₂ to euthanize mice require further investigation. We have highlighted some of the areas that we believe require more in-depth analysis to fill the gaps in our understanding of CO₂ euthanasia.

Sex

Studies investigating CO₂ euthanasia have largely used a single sex, and comprehensive comparison studies involving both males and females are infrequent. The few studies that used both sexes did not find any significant difference between males and females during CO₂ euthanasia. Authors who used male (n = 18) and female (n = 30) mice found no differences between sexes in LORR, blood gas analysis, or behavior in a study that investigated combination gas euthanasia with CO₂ and nitrogen.¹⁰² In addition, sex-associated differences in plasma norepinephrine in rats (n = 10 per group) were inapparent in a CO₂ euthanasia study.¹⁵

Strain

Rodent strain- and sex-associated differences influence HPA response to stressors,^52,60,85 thus highlighting the importance of understanding strain- or sex-related behavioral or physiologic differences to CO₂ euthanasia. In one study, strain-associated differences were noted in the time to LORR as well as in jumping and rearing behaviors: C57BL/6 mice had a shorter time to LORR and jumped and reared more frequently than did CD1 mice during CO₂ euthanasia.¹⁰² Authors that compared pulmonary hemorrhage in 6-wk-old and 6-mo-old male BALB/c mice and C57BL/6 mice found that age had no bearing on pulmonary hemorrhage severity, but severity was markedly increased in BALB/c compared with C57BL/6 mice at both 21% CO₂ CRR and with a CO₂-prefilled chamber.³⁰ In addition, BALB/c mice euthanized with isoflurane did not have pulmonary hemorrhage.³⁰ Thus, there is evidence for differing responses between different strains of mice in several variables, and these differences should be considered when selecting a euthanasia method.

Group housing

Awareness regarding the potential that rodents are experiencing and displaying empathy is increasing.^{22,35,49,59,87,88} Empathy refers to an animal's ability to experience the feelings of another animal. Emotional contagion is the simplest form of empathy and is exhibited, for example, by the presence of a stress reaction in one animal when another is in pain. In one study, mice cohoused for 14 or 21 d with another mouse demonstrated empathy, yet did not respond to a mouse that was a stranger.⁵⁹ Similarly, foot-shocking a cage mate elicited freezing behavior in the observer, but shocking strange mice did not elicit this behavior.³² These studies establish a firm basis for the assertion that mice exhibit empathy and for the importance of familiarity during this exposure. Consequently, whether mice should be euthanized with other mice in the cage and whether it is important that the mice are familiar with those being euthanized simultaneously should be addressed, and further research in this area is warranted.

Conclusions

The 2006 and 2013 Newcastle reports both concluded that alternatives to the use of CO₂ should be identified. Several studies indicate that the use of an alternative gas, such as isoflurane, is not a humane improvement in terms of euthanasia. We do not support a mandatory requirement to anesthetize mice with isoflurane prior to CO₂ euthanasia. In addition, the use of alternative methods such as argon and nitrogen are not improvements in terms of euthanasia.

The AVMA Panel on Euthanasia recommends the use of CO₂ at 10% to 30% CRR. The Canadian Council on Animal Care Guidelines on Euthanasia of Animals Used in Science and attendees at the 2006 and 2013 Newcastle meeting on euthanasia recommended the use of anesthetics prior to CO₂ euthanasia.^19,37,38 Given the results of our review, we disagree with those conclusions. As long as animals do not experience pain (mice are unconscious before CO₂ reaches the pain-inducing level of 40%), we recommend using faster CO₂ CRR to decrease the time mice are experiencing distress.

Euthanasia should minimize pain and distress. According to current knowledge, the recommended use of CO₂ does not lead to pain. Although stress is present during the euthanasia process with CO₂, all euthanasia procedures available currently lead to an element of stress. Therefore, in the absence of a better alternative agent, we recommend the continued humane use of CO₂ for the euthanasia of laboratory rats and mice. In addition, we concur with the Newcastle reports that additional studies are required. Please note these recommendations are specific for laboratory rats and mice and cannot be extrapolated to other rodent species, given that the behaviors of these species may differ significantly.