SYSTEMATIC REVIEW published: 22 July 2020 doi: 10.3389/fvets.2020.00411 Welfare Impact of Carbon Dioxide Euthanasia on Laboratory Mice and Rats: A Systematic Review Patricia V. Turner 1,2*, Debra L. Hickman 3 , Judith van Luijk 4 , Merel Ritskes-Hoitinga 4 , Jan M. Sargeant 5,6 , T. Miki Kurosawa 7 , Takashi Agui 8 , Vera Baumans 9 , Woo Sung Choi 10 , Yang-Kyu Choi 11 , Paul A. Flecknell 12 , Byeong H. Lee 13 , Pedro J. Otaegui 14 , Kathleen R. Pritchett-Corning 15,16 and Keisuke Shimada 17 1 Department of Pathobiology, University of Guelph, Guelph, ON, Canada, 2 Global Animal Welfare and Training, Charles River, Wilmington, MA, United States, 3 Laboratory Animal Resource Center, School of Medicine, Indiana University, Indianapolis, IN, United States, 4 Department of Health Evidence, SYstematic Review Center for Laboratory Experimentation (SYRCLE), Radboud University, Nijmegen, Netherlands, 5 Department of Population Medicine, University of Guelph, Guelph, ON, Canada, 6 Centre for Public Health and Zoonoses, University of Guelph, Guelph, ON, Canada, 7 Faculty of Veterinary Medicine, Kagoshima University, Kagoshima, Japan, 8 Department of Applied Veterinary Science, Faculty of Veterinary Medicine, Hokkaido University, Sapporo, Japan, 9 Department of Animals, Science and Society, Utrecht University, Utrecht, Edited by: Netherlands, 10 National New Drug Development Cluster, Woojung Bio, Suwon-si, South Korea, 11 Department of Laboratory Emily Patterson-Kane, Animal Medicine, College of Veterinary Medicine, Konkuk University, Seoul, South Korea, 12 Institute of Neuroscience, National Coalition of Independent Newcastle University, Newcastle upon Tyne, United Kingdom, 13 Osong Medical Innovation Foundation, Cheongju, Scholars, United States South Korea, 14 Laboratory Animal Facilities, Autonomous University of Barcelona, Barcelona, Spain, 15 Office of Animal Reviewed by: Resources, Harvard University Faculty of Arts and Sciences, Cambridge, MA, United States, 16 Department of Comparative Yvonne A. Dzal, Medicine, University of Washington, Seattle, WA, United States, 17 Animal Resource Center for Infectious Diseases, Research University of Winnipeg, Canada Institute for Microbial Diseases, Osaka University, Osaka, Japan Jeremy Davidson Bailoo, Texas Tech University Health Sciences Center, United States Background: There has been increased concern about the suitability of CO2 as a Megan Renee LaFollette, method for euthanasia of laboratory mice and rats, including the potential discomfort, The North American 3Rs Collaborative, United States pain or distress that animals may experience prior to loss of consciousness; time to loss *Correspondence: of consciousness; best methods for use of CO2 ; and the availability of better alternatives. Patricia V. Turner These discussions have been useful in providing new information, but have resulted in pvturner@uoguelph.ca significant confusion regarding the acceptability of CO2 for rodent euthanasia. In some Specialty section: cases, researchers and veterinarians have become uncertain as to which techniques to This article was submitted to recommend or use for euthanasia of laboratory mice and rats. Animal Behavior and Welfare, a section of the journal Methods: The International Association of Colleges of Laboratory Animal Medicine Frontiers in Veterinary Science (IACLAM) convened a taskforce to examine the evidence for adverse welfare indicators Received: 13 April 2020 in laboratory rats and mice undergoing CO2 euthanasia using a SYRCLE-registered Accepted: 09 June 2020 systematic review protocol. Of 3,772 papers identified through a database search Published: 22 July 2020 (PubMed, Web of Science, CAB Direct, Agricola, and grey literature) from 1900 to 2017, Citation: Turner PV, Hickman DL, van Luijk J, 37 studies were identified for detailed review (some including more than one species or Ritskes-Hoitinga M, Sargeant JM, age group), including 15 in adult mice, 21 in adult rats, and 5 in neonates of both species. Kurosawa TM, Agui T, Baumans V, Choi WS, Choi Y-K, Flecknell PA, Experiments or reports were excluded if they only assessed parameters other than those Lee BH, Otaegui PJ, directly affecting animal welfare during CO2 induction and/or euthanasia. Pritchett-Corning KR and Shimada K (2020) Welfare Impact of Carbon Results: Study design and outcome measures were highly variable and there was an Dioxide Euthanasia on Laboratory unclear to high risk of bias in many of the published studies. Changes in the outcome Mice and Rats: A Systematic Review. Front. Vet. Sci. 7:411. measures evaluated were inconsistent or poorly differentiated. It is likely that repeated doi: 10.3389/fvets.2020.00411 exposures to carbon dioxide inhalation are aversive to adult rats and mice, based on Frontiers in Veterinary Science | www.frontiersin.org 1 July 2020 | Volume 7 | Article 411 Turner et al. CO2 Euthanasia of Laboratory Rodents avoidance behavior studies; however, this effect is largely indistinguishable from aversion induced by repeated exposures to other inhalant anesthetic gasses. Conclusion: There is insufficient evidence to permit an unbiased assessment of the effect of CO2 inhalation during euthanasia on welfare indicators in laboratory mice and rats. Additional well-designed, unbiased, and adequately powered studies are needed to accurately assess the welfare of laboratory mice and rats undergoing euthanasia via CO2 gas. Keywords: carbon dioxide, euthanasia, systematic review, mouse, rat, animal welfare, pain, distress INTRODUCTION ∼15–20%. Prolonged exposure to high levels of CO2 (∼40– 50%) results in coma, apnea, hyperkalemia, and cardiac arrest Euthanasia, or provision of a good death, is considered a critical (6, 7). While its use as an anesthetic agent in rodents has event in an animal’s life and it is important for animal well- largely been discontinued for over two decades, CO2 gas being that it be conducted in a humane manner. According inhalation is used widely around the world for research to international guidelines on research animal euthanasia, a rodent euthanasia. euthanasia procedure should result in rapid and irreversible Recently, there has been concern that CO2 inhalation is loss of animal consciousness with minimal distress leading to inhumane as a method of euthanasia for laboratory rodents eventual death (1). It should also be relatively easy, inexpensive, (8–10). Carbon dioxide is a normal component of inhaled air and safe for a trained operator to perform, be accessible to (0.04%) and internal sensors for CO2 levels drive respiration in a wide range of possible users, and be esthetically acceptable most animals. Labored breathing and fear have been noted in to those performing or observing the procedure (1). Mice and mice when inhaled CO2 levels are ∼5–15%. Fear is thought to be rats commonly worked with in science and there has been elicited via stimulation of an acid-sensing ion channel 1a located significant interest in recent years in determining the best in the amygdala (11). Pain is thought to occur when conscious method(s) for euthanizing these animals in research settings (1). animals are exposed to high levels of CO2 (>47%) because of the Humane methods are required for euthanizing infant rodents, carbonic anhydrase found within mucosal surfaces of the upper juveniles, adults, and pregnant dams—as individual animals, in and lower respiratory tree, which converts CO2 to carbonic acid small groups, and sometimes in very large numbers. Research in the presence of water [reviewed in (9)]. This suggests that there rodents are also euthanized because of spontaneous pain, is a potential for discomfort, distress, and/or pain to occur in sickness, injury, or deformity; because humane or experimental mice and rats prior to the loss of consciousness when CO2 gas is endpoints are reached; as a means of population management; used for euthanasia. However, there is conflicting information in and because of environmental emergencies (e.g., flooding or published studies regarding the effects of CO2 gas inhalation on power failures) or depopulation needs (e.g., a biosecurity mice and rats during induction for euthanasia. This has resulted break). Given the wide variation of reasons for euthanasia, in uncertainty regarding the acceptability of inhaled CO2 for the methods must be safe for a range of different situations. rodent euthanasia and uncertainty by veterinarians and others, Additionally, because animals are often euthanized at study end such as researchers, regarding which techniques to recommend to harvest organs for additional investigations, the method of or use for euthanasia of laboratory mice and rats. euthanasia should not interfere with the future use of tissues. Because euthanasia is deemed a critical responsibility of A further consideration for euthanasia method is that it should the laboratory animal veterinarian (1, 12), the International not create any hazards for carcass disposal. For example, if Association of Colleges of Laboratory Animal Medicine cadavers are to be donated for feeding of zoo animals or (IACLAM; iaclam.org) identified a need to review the literature captive wildlife then the cadavers must be free of substances objectively to determine whether CO2 gas inhalation meets the that may adversely impact other animals. For all of these definition of euthanasia (1), including evaluating criteria such reasons, carbon dioxide (CO2 ) inhalation is the most common as time to loss of consciousness and time to death; whether technique in use today for euthanasia of laboratory mice and the method is suited to one or multiple animals; and further to rats (2, 3). compare cost, availability, operator safety, practicality of use in Narcosis following carbon dioxide gas exposure has been long different circumstances, and esthetics of CO2 use on operators recognized, and CO2 gas was used historically as a short term or observers. A task force of board certified laboratory animal anesthetic agent for humans and animals for almost 200 years veterinarians from around the world was convened to conduct a (3–5). Increasing concentrations of inhaled CO2 (hypercarbia) systematic review of the welfare impact (i.e., any behavioral or induces respiratory and cerebrospinal fluid acidosis, as well physiologic effects related to distress, aversion, and/or pain) of as cardiovascular depression, that eventually lead to stupor, exposure to CO2 gas (alone or in combination with other agents) with unconsciousness occurring at CO2 concentrations of for euthanasia of neonatal and adult laboratory mice and rats. Frontiers in Veterinary Science | www.frontiersin.org 2 July 2020 | Volume 7 | Article 411 Turner et al. CO2 Euthanasia of Laboratory Rodents MATERIALS AND METHODS Data Extraction and Data Analysis Data from studies meeting the study selection criteria were Study Design and Systematic Review independently extracted by two reviewers (PT and DH) using a Protocol standardized form, which was tested on 3 pre-selected studies A systematic review of the literature was conducted and reported (JS). Discrepancies in data extraction were resolved by consensus. using the PRISMA guidelines for reporting of systematic reviews Study level data included year published, country of origin [(13) and Supplementary Table 1]. The review protocol was of the work, authorship, and funding source. Population evaluated and registered at http://www.syrcle.nl on May 3, characteristics included rodent species (mouse, rat, or both), 2017 (14). sex (only males, only females, or mixed sex populations), age group of rodents being studied (neonatal vs. post-weaned), stock or strain of the rodent models being evaluated, and whether Search Strategy and Data Sources animals had been used for gas exposure or anesthesia studies On August 1, 2017, a systematic literature search was conducted in the past. Study design characteristics included the number concerning CO2 euthanasia of mice and rats using the following of experimental and control groups, the number of animals electronic databases: Medline (PubMed), Web of Science, CAB per group, and whether studies were single exposure euthanasia Direct, and Agricola (see Supplementary Table 2). No language experiments or multiple exposure “anesthesia with recovery” or date restrictions were placed aside from database onset experiments. Intervention characteristics included gas flow rates, dates (Medline, 1950; Web of Science, 1900; CAB Direct, 1904; duration of exposure, type of gas or euthanasia agent applied, Agricola, 1970). A research librarian from the University of concentration or ratio of gas in mixtures, and other euthanasia Guelph was consulted on the search strategy. In addition, treatments. Only quantifiable outcome measures were included, reference lists were checked from all identified review articles and, where possible, mean (±SE) time of occurrence, number of on euthanasia of laboratory rodents as well as included studies occurrences, or change in the outcome measure were collected. for possible relevant references. An English-only grey (i.e., non- If a study described pilot data together with definitive data only peer reviewed) literature search using Google Scholar was also outcomes from the definitive studies were collected. Continuous performed for identifying possible graduate theses, study papers, outcome measures specifically collected included: heart rate, and abstracts. blood pressure, plasma corticosterone levels, time to loss of posture, time to loss of righting reflex, time to death, time to onset of labored breathing, and change in EEG activity, and dichotomous outcome measures collected included: increased c- Study Selection, Interventions, and fos expression in the brain, urination, defecation, vocalization, Comparators seizures, escape behaviors, increased activity on induction, and Primary reviewers (PT, DH, and TK) are board certified bleeding from the nose during induction. Individual studies laboratory animal veterinarians and members of the IACLAM varied in the definitions and methods used for many of these Taskforce on CO2 Euthanasia. For Phase 1 screening, one variables. Data were extracted from tables or graphs, if numerical reviewer (PT) initially pre-screened all references based on data was not reported. title alone to remove obvious irrelevant references and all excluded reference titles were checked by two individuals (DH Risk of Bias and Quality Assessment and TK). In cases of conflict, the title was retained for more Assessment of risk of bias and study quality were conducted detailed title and abstract screening. For Phase 2 screening, independently by two reviewers (PT and DH) using the modified each reference was assessed by two independent reviewers SYRCLE Risk of Bias tool (15) and any disagreements were [PT and (DH or TK)] based on the title and abstract with resolved by consensus. The risk of bias tool was originally consensus agreement achieved for all retained papers. Studies developed for studies with separate control and intervention were excluded if they were not an in vivo intervention (i.e., groups. Only one study used separate non-treatment and experimental or observational) study in mice or rats of any comparison groups, thus risk of bias was evaluated for all age, if CO2 inhalation or another inhalant-type anesthesia or single trial euthanasia-only studies in which there were two or euthanasia method was not evaluated, and if publications were more comparison groups. Reporting of animal randomization, from conference proceedings or only available in abstract form blinding, and sample size calculations was also assessed for all and would not permit further data extraction or assessment of reviewed papers as indicators of study quality. risk of bias. Remaining papers were imported to EndNoteX7 (Clarivate Analytics, Philadelphia, PA, USA) and duplicates were RESULTS removed. For Phase 3 screening, each full-text reference was assessed (PT and DH) using an online systematic review program Study Selection and Characteristics (Distiller SR, Evidence Partners Inc., Ottawa, ON, Canada), and Results of the search strategy and study selection are presented all conflicts were resolved by consensus. Studies were excluded if in Figure 1. Of the 108 full text articles reviewed, 71 articles they met any of the original three exclusion criteria and/or the were excluded as they did not meet eligibility criteria and 37 study did not measure any outcomes directly relevant to animal articles were included. In total, 15 papers were reviewed for welfare or behavior. mice (16–30), 21 papers for rats (16, 22, 23, 31–48), and 5 Frontiers in Veterinary Science | www.frontiersin.org 3 July 2020 | Volume 7 | Article 411 Turner et al. CO2 Euthanasia of Laboratory Rodents FIGURE 1 | Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) study flow diagram [modified from (13)]. papers were reviewed for neonatal rodents (21, 49–52), with induction with CO2 or other inhalant gases and then recovered three studies reporting results for both adult mice and rats animals for repeated exposures. Detailed study characteristics (16, 22, 23) and one study reporting results for both adult and findings are presented in Tables 1–3 for adult mice, adult female mice and their pups (21, 49–52). Not all studies evaluated rats, and neonatal rodents, respectively. euthanasia since welfare concerns in using inhalant gasses for There was considerable diversity in the methods used for euthanasia primarily revolve around what the animal experiences exposure to CO2 and in all but one study, there was no attempt during the period from onset of exposure to the gas until loss to use a separate control or sham-treated group that wasn’t later of consciousness. Once an animal is deeply anesthetized, it is exposed to a secondary euthanasia or anesthesia intervention by definition insensible to pain and distress as long as the within the same study. In no case did three or more studies animal does not recover. Thus, some of the studies included in from different laboratories use similar interventions, species, this review evaluated rodent behaviors, physiology, and time to and outcome measures, or a separate control group, precluding Frontiers in Veterinary Science | www.frontiersin.org 4 July 2020 | Volume 7 | Article 411 Turner et al. CO2 Euthanasia of Laboratory Rodents TABLE 1 | Characteristics of the 15 adult mouse studies included in the systematic review. References Strain, Sex, and Intervention Euthanasia or Frequency of Control intervention Outcome measures Age treatments of anesthesia exposure or (group size) (identification of effect) interest (group induction with re-use size) recovery (16) Quackenbush, 1. sublimation of dry Euthanasia None, three Untreated controls in same Chamber escape behaviors: females, mature ice replicates environment in two ND 2. Chloroform into densities, n = 1 and 3 LOP: shorter for CO2 bedding Three replicates for each Time to death: shorter for 3. Ether-soaked group CO2 cotton (n = 12) (17) C57BL/6N, males, 1. CO2 15% VDR Definitive: Re-used control Pilot: air flow Anxiety: ND 16 week 2. CO2 30% VDR euthanasia mice in euthanasia Controls: noise and air Heart rate: ND between 3. CO2 50% VDR studies movement-only exposure, groups; blood pressure: ND 4. CO2 100% VDR n = 28 LOP: faster as [CO2 ] (n = 12–13) increased Plasma cort: ND Time to death: decreased as [CO2 ] increased (18) C57BL/6NTac, 1. Isoflurane 5% in Euthanasia None Used same control and CO2 Anxiety: increased as [CO2 ] males, 16 week O2 1 L/min then comparator data as increased distance traveled: 100% CO2 after published in (17) increased as [CO2 ] induction increased 2. Pentobarbital IP Heart rate: ND or increased (n = 11–14) as [CO2 ] increased Blood pressure: ND or decreased as [CO2 ] increased Plasma ACTH: reduced with pentobarb LOP: ND or faster as [CO2 ] increased Time to death: ND or faster as [CO2 ] increased (19) 129×S1/SVJ, mixed 1. CO2 100% euthanasia under None Each animal used as own Time to loss of EEG signal: sex, 6–24 week facemask succinyl choline control for baseline data shorter for CO2 100% = 2. CO2 70% KCl = cervical dislocation = facemask decapitation 3. KCl Visual evoked potential: 4. cervical reduced for CO2 100% dislocation 5. decapitation (n = 5–10) (20) NMRI, males, 12–16 1. CO2 0.2 L/min FR Gas exposure with 4 days chamber Air only central chamber, Approach/avoidance with week 2. 8% sevoflurane in recovery habituation, no gas with two lateral chambers food reward: time in sevo O2 0.2 L/min FR into which gas pumped, chamber increased 3. 4% isoflurane in food restriction to maintain compared with air only, ND O2 0.2 L/min FR 85–95% pre-study weight, between sevo vs. CO2 (n = 6–9) food pellets placed in lateral chambers, ND between chambers during testing to time in iso vs. CO2 motivate entry, compared chambers, ND between combinations of gas time in iso vs. sevo exposures in three trials: chambers, ND in food sevo vs. CO2 , CO2 vs. iso, reward eaten between three iso vs. sevo chambers (21) Mixed strains, cull 1. CO2 20% VDR Euthanasia or None None, all compared to CO2 LOP: not consistently genotypes, females, 2. Pentobarbital IP anesthesia induction measured mature, and 3. Cervical with euthanasia time to cardiac arrest: pregnant dislocation fastest KCl, next fastest 4. Halothane in O2 CO2 = cervical dislocation 5. KCl under (with or without anesthesia) anesthesia 6. Cervical dislocation under anesthesia (n = 5–7) (Continued) Frontiers in Veterinary Science | www.frontiersin.org 5 July 2020 | Volume 7 | Article 411 Turner et al. CO2 Euthanasia of Laboratory Rodents TABLE 1 | Continued References Strain, Sex, and Intervention Euthanasia or Frequency of Control intervention Outcome measures Age treatments of anesthesia exposure or (group size) (identification of effect) interest (group induction with re-use size) recovery (22) BALB/C, females, 1. CO2 53% FR Definitive: gas Yes, up to 4 Pilot: gas level testing at low, Rearing: increased for CO2 9–15 weekk 2. argon 99% FR exposure with exposures per medium, high, n = 30 (not 53% FR from baseline, ND 3. CO2 20%/argon recovery mouse over 6 week re-used in definitive study) between treatment groups 4.5% FR study Definitive: 30 min chamber approach/avoidance (time 4. CO2 30%/argon acclimation period, air-only dwelling and time to 5.3% FR (n = 30 3 min control (each animal withdrawal latencies): all total) used as own control), [CO2 ] shorter vs. argon -high concentrations measured time dwelling, alone only time to withdrawal, and time to re-entry in chamber (23) BALB/C, females, 1. CO2 50.8% FR Definitive: gas Up to 7 exposures Pilot: gas level testing at low, Anxiety/pain behaviors 9–15 week 2. CO2 34.9% FR exposure with per mouse medium, high, n = 30 (not (urination, defecation, 3. CO2 25.5% FR recovery re-used in definitive study) rearing): ND from baseline, 4. Argon 99.2% FR Definitive: 30 min chamber ND between groups (n = 30 total) acclimation period, air-only Approach avoidance -high concentrations 3 min control (each animal latencies: time dwelling: only used as own control), shorter for all compared measured time dwelling, with air only, ND between time to withdrawal, and time groups; time to withdrawal to re-entry in chamber latency: ND (24) CD1, males, surplus 1. CO2 70% FR Gas exposure with >50 exposures per Control: air only, each Approach/avoidance: mature 2. AR 160% FR recovery mouse over 4 month animal as its own control latency to withdrawal 3. CO 9% FR study, three -initial training to lower decreased with increasing 4. isoflurane 3%/O2 replicates per chamber with food reward [CO2 ] (always left when 70% FR concentration -multiple CO2 chamber concentrations (n = 6–7) concentrations evaluated 13.5–18.2%), latency to withdraw for isoflurane = CO, latency to withdraw longest for argon (25) C57BL/6J-Tyr, 1. CO2 50% FR Euthanasia None None Labored breathing: seen in females, 20 week 2. CO2 40% FR all groups prior to LOP, ND 3. CO2 30% FR seen between groups 4. CO2 20% FR (n = LOP: faster with increased 5–6) [CO2 ] LORR: faster with increased [CO2 ] (26) C57BL/6J, males, 8 1. CO2 20% VDR Gas exposure with Two replicates Three acclimation trials to Light/dark aversion pairing week 2. isoflurane 5%/O2 recovery test apparatus chamber latency: time to 4 L/min withdrawal: iso/O2 > CO2 = 3. isoflurane 5% isoflurane drop, re-entry drop (n = 8–9) time: iso/O2 < CO2 = isoflurane drop; re-exposure: time to withdrawal reduced for iso/O2 (27) C57BL/6J, males, 1. CO2 20% VDR Euthanasia None None LOP: ND 12–16 week 2. isoflurane 5% in LORR: faster with CO2 O2 2 L/min (n = time to death: faster with 6–7/group) CO2 (28) CD1, retired 1. CO2 70% FR, Euthanasia None, three Chamber light levels during Anxiety combined: highest sentinels, sex, and 325 lux replicates euthanasia evaluated, for CO2 70% = CO2 70% + age not specified 2. CO2 30% FR, euthanized in pairs in home iso, increased at 500 vs. 40 325 lux cages, no acclimation lux = CO2 30% 325 lux 3. isoflurane 3% period for lighting prior to LOP: CO2 70% < CO2 70% then 70% CO2 FR euthanasia + iso < CO2 30% 325 lux after induction, 325 Plasma cort: increased for lux CO2 70% + iso 4. CO2 30% FR, 500 lux 5. CO2 30% FR, 5 lux (n = 6) (Continued) Frontiers in Veterinary Science | www.frontiersin.org 6 July 2020 | Volume 7 | Article 411 Turner et al. CO2 Euthanasia of Laboratory Rodents TABLE 1 | Continued References Strain, Sex, and Intervention Euthanasia or Frequency of Control intervention Outcome measures Age treatments of anesthesia exposure or (group size) (identification of effect) interest (group induction with re-use size) recovery (29) CD1, mixed sex, 1. CO2 20% VDR Definitive: Pilot: 3× with 48 h Pilot: exposure with Rearing: ND 10–12 week 2. CO2 20% + 60% euthanasia washout between recovery, (n = 12 F) Jumping: increased for CO2 N2 O VDR trial + N2 3. 20% CO2 + 60% LORR: fastest for CO2 + N2 VDR (n = N2 O 3/sex/strain) (30) CD1, females, 8–11 1. CO2 100% VDR Euthanasia None Exposure to CO2 gas for Increased activity: higher for week 2. CO2 20% VDR euthanasia with or without iso 3. iso 5% in O2 1.2 sedation Ultrasonic vocalization: L/min then CO2 increased for iso = CO2 100% VDR after 100% VDR induction Other behaviors: ND 4. CO2 20% VDR + LORR: fastest for CO2 ACE premed 100% VDR 5. CO2 20% VDR + Plasma corticosterone: MDZ premed (n = increased MDZ plasma 10) ACTH: ND c-fos: decreased with CO2 100% VDR Time to death: fastest for CO2 100% VDR ACE, acepromazine; ACTH, adrenocorticotropin hormone; cort, corticosterone; EEG, electroencephalograph; FR, flow rate; IP, intraperitoneal; iso, isoflurane; KCl, potassium chloride; LOP, time to loss of posture; LORR, time to loss of righting reflex; MDZ, midazolam; ND, no difference; sevo, sevoflurane VDR, volume displacement rate. combination of data for meta-analyses. An attempt to re- had been completed, such that no randomization or blinding analyze outcome data was similarly unsuccessful because of the could occur for this group. Most studies (9 of 16) did not provide heterogeneity in interventions, outcome measures collected, and information about blinding the treatments provided to animals their timing. or blinding those assessing the treatment effects. Because of study designs, treatments could not be blinded in 4 of 16 studies. Risk of Bias and Quality Assessment For this reason, we interpreted both the risk of performance Nine of 15 mouse studies (16–19, 21, 25, 27, 28, 30), 7 of bias and the risk of detection bias to be moderate to high 21 rat studies (16, 32, 34, 35, 37, 39, 47), and two of five for most studies. No information about study drop-outs was rodent pup studies (21, 49) (n = 16 separate studies as two provided in 10 of 16 studies, and this lack of reporting was evaluated both mice and rats of mice of different ages) were deemed a moderate to high risk in 3 of 16 studies, in which defined as single trial euthanasia studies and were assessed for study drop-outs were noted but not discussed further. In 7 of 16 risk of bias, although none included separate untreated and studies, some results or methods were not fully reported. This comparison groups. Other studies that were part of this review included inconsistent reporting of results or methods, reporting had incompatible study designs for the risk of bias assessment of results in figures only (with or without error bars or confidence (i.e., single observation group, crossover studies, animals used in intervals), and not accounting for re-use of animals between other gassing experiments, animals used in gas pilot studies prior different phases of the study. In 8 of 16 studies, a moderate to the definitive study, animals exposed to repeated subanesthetic to high risk for carry-over effects was identified due to re-use gas concentrations or used repeatedly in induction-only trials). of animals between different phases of the study or use of a The results of the risk of bias assessment are summarized cross-over design. in Figure 2. In general, bias reduction measures were poorly The overall quality assessment for the 37 papers reported by authors and were assessed as “no information.” In 11 included in this systematic review is shown in Figure 3. of 16 studies, no information was provided about randomization In general, randomization of animals to study groups of animals to different comparison groups. In the other five was reported in 12 of 37 studies (32%), although, as studies, randomization was mentioned; however only one of mentioned above, only one paper provided actual evidence the studies provided actual evidence of randomization of some of randomization. Reporting of blinding for data collection of their treatment groups. Sample size or power calculations or assessment occurred in 10 of 37 studies (27%), and were only mentioned in 1 of these 16 studies. In half of the reporting of a sample size calculation occurred in 4 of 37 16 studies, animals served as their own controls; however, no studies (11%). information was provided in 5 of 16 studies, and in one study, a The risk of confirmation bias was determined to be high for treatment group was added after all other portions of the study some of the rat studies in which the same rats were re-used Frontiers in Veterinary Science | www.frontiersin.org 7 July 2020 | Volume 7 | Article 411 Turner et al. CO2 Euthanasia of Laboratory Rodents TABLE 2 | Characteristics of the 21 adult rat studies included in the systematic review. References Strain, Sex, and Intervention Euthanasia or Frequency of Control intervention Outcome measure Age treatments of anesthetic exposure or (group size) (identification of effect) interest (group induction with re-exposure size) recovery (16) Wistar, mixed sex, 1. Sublimation of dry Euthanasia None, three Untreated controls in same Escape behaviors: ND mature 220 g, and ice replicates environment in 2 densities, Facial rubbing in young rats immature 88 g 2. Chloroform into n = 1 and 3 (two for ether) increased for ether bedding -three replicates for each LOP: fastest for CO2 3. Ether-soaked Time to death: fastest for cotton (n = 11 adult, CO2 n = 12 young) (31) SD, males, 12–24 1. Isoflurane/O2 Anesthesia induction Aversion-avoidance Used training sessions with Aversion-avoidance week 2. Sevoflurane/O2 with recovery (light-dark), 3 times O2 only as control latencies: ND, both aversive (n = 17 + training and increased with aversion-avoidance, Approach re-exposure n = 9 approach avoidance, no. of Approach-avoidance avoidance) training trials not latencies: ND, both aversive specified and increased with re-exposure (32) SD, males, 8–16 1. CO2 10% VDR Euthanasia None Control: each animal used Rearing, sniffing, and week 2. Argon 50% VDR as its own control with O2 ambulating: ND between (n = 8) only for baseline data groups or from baseline gasping and seizures: increased for argon only Ultrasonic vocalization: none recorded for any group Heart rate prior to LOP: faster for CO2 LOP: fastest for argon Time to death: fastest for argon (33) SD, females, 7–9 1. CO2 30% VDR Repeated gas Not specified Control: subset of animals 51 KHz ultrasonic week 2. isoflurane exposure with exposed to O2 only (n = 6) vocalizations: increased as 2.5%/O2 (nine recovery [CO2 ] increased, no animals total, n = 3 vocalizations with isoflurane CO2 only, n = 6 isoflurane and O2 ) (34) SD, females, 1. CO2 20% VDR Euthanasia None None LOP: ND 250–500 g 2. isoflurane 5%/O2 LORR: fastest for 20% CO2 then 100% CO2 Heart rate: decreased as 3. CO2 /O2 [CO2 ] increased 70%/30% Electromyelogram: ND 4. pentobarbital Time to isoelectric (n = 8) electrocortigraph: faster with 20% CO2 Time to death: fastest for 20% CO2 (35) Wistar, males, 1. CO2 100% prefill Euthanasia Used in sleep/wake None Gasping: increased with 32–52 week 2. CO2 22.5 L/min study previously, 100% CO2 prefill 3. CO2 2.5 L/min details not provided LOP: fastest for 100% CO2 4. CO2 22.5 prefill L/min/O2 11.25 Heart rate: decreased with L/min (n = 7) 100% CO2 prefill and fast flow CO2 Time to isoelectric EEG: fastest for 100% CO2 prefill (36) SD (3 different 1. CO2 100% prefill Definitive: Used in CO2 Pilot: anesthesia Nasal bleeding and sources), sex not 2. CO2 80% euthanasia anesthesia pilot salivation prior to death: specified, 3. CO2 70% study increased with 60% and 300–1,050 g 4. CO2 60% n = 6 in 70% CO2 each of precharged LOP: ND and empty trials at LORR: faster as [CO2 ] each concentration increased Time to death: faster as [CO2 ] increased (Continued) Frontiers in Veterinary Science | www.frontiersin.org 8 July 2020 | Volume 7 | Article 411 Turner et al. CO2 Euthanasia of Laboratory Rodents TABLE 2 | Continued References Strain, Sex, and Intervention Euthanasia or Frequency of Control intervention Outcome measure Age treatments of anesthetic exposure or (group size) (identification of effect) interest (group induction with re-exposure size) recovery (37) F344, males, 8 week 1. CO2 6 L/min Euthanasia None None, groups of animals No urination, defection, 2. Ace 7 mg + CO2 sequentially killed by vocalization or signs of pain 6 L/min decapitation at three in any group 3. Pentobarb 60 timepoints (30, 75, 120 s) Increased tachypnea: all mg/kg + CO2 6 groups L/min (n = 4) Serum glucose: ND Plasma [ACTH]: higher for pentobarb plasma [cort]: higher for pentobarb (38) SD, males, 32 week 1. CO2 100% prefill, Definitive: Used in CO2 Pilot: anesthesia No distress observed for catheterized euthanasia with anesthesia pilot any group 2. CO2 7.2 L/min blood gas collection study and then LOP: faster for 100% CO2 FR, catheterized rats used in previous carotid artery prefill (n = 8) behavioral study, catheter placed LORR: faster for 100% CO2 details not provided Under anesthesia for prefill definitive phase Time to death: faster for 100% CO2 prefill (39) SD, males, 1. CO2 100% VDR Euthanasia, None Each animal used as its own No seizures noted in any 300–350 g 2. CO2 30% VDR radiotelemetry control with air only for animals gasping: increased 3. CO2 10% VDR implantation under baseline data for first 20 s with 100% CO2 (n = 7) isoflurane anesthesia Locomotor activity: ND from baseline LOP: faster for 100% CO2 Heart rate: decreased for all groups with time Blood pressure: decreased for all groups with time Time to death: faster for 100% CO2 (40) Wistar, males, 40 1. CO2 14.5% VDR Anesthesia Eight times in this Control: each animal used Approach/avoidance: all gas week 2. CO2 70%/ O2 study with four as its own control with air treatments reduced latency 30% 14% VDR training trials only for baseline data -initial to stop eating, latency to 3. CO2 70%/O2 Rats had been used training to lower chamber leave chamber, and number 30% 21% VDR multiple times in with food reward of food rewards eaten 4. Air (n = 8) previous CO2 compared with air anesthesia studies, details not provided (22) Wistar, females, 1. CO2 53% FR Definitive: gas Yes, up to four Pilot: gas level testing at low, Rearing behavior: ND 9–15 week 2. argon 99% FR exposure with exposures per rat medium, high, n = 30 (not between treatment groups 3. CO2 20%/argon recovery over 6 week study re-used in definitive study) and baseline grooming face 4.5% FR Definitive: 30 min chamber and sniffing: increased from 4. CO2 30%/argon acclimation period, air-only baseline variably 5.3% FR (n = 30 3 min control (each animal approach/avoidance total) used as own control), latencies time dwelling and -high concentrations measured time dwelling, time to withdrawal: shorter only time to withdrawal, and time for all [CO2 ] vs. argon only to re-entry in chamber (23) Wistar, females, 15 1. CO2 50.8% FR Definitive: gas Up to seven Pilot: gas level testing at low, Anxiety/pain behaviors week 2. CO2 34.9% FR exposure with exposures per rat medium, high, n = 30 (not (urination, defecation, 3. CO2 25.5% FR recovery re-used in definitive study) rearing): ND from baseline, 4. Argon 99.2% FR Definitive: 30 min chamber ND between groups (n = 30 total) acclimation period, air-only Approach avoidance: -high concentrations 3 min control (each animal latency time dwelling: only used as own control), shorter for all compared with measured time dwelling, air only, ND between groups time to withdrawal, and time to re-entry in chamber (Continued) Frontiers in Veterinary Science | www.frontiersin.org 9 July 2020 | Volume 7 | Article 411 Turner et al. CO2 Euthanasia of Laboratory Rodents TABLE 2 | Continued References Strain, Sex, and Intervention Euthanasia or Frequency of Control intervention Outcome measure Age treatments of anesthetic exposure or (group size) (identification of effect) interest (group induction with re-exposure size) recovery (41) Wistar, males, 44 1. Ar 239% Definitive: gas Multiple times in two Control: 32 control air trials Approach avoidance: week 2. Ar 199% exposure with phases of this study (same animals) number of cereal pieces 3. Ar 159% recovery with air training trials eaten + latency to leave 4. Ar 120% (n = 7) (details not provided) shorter with increased [Ar] Rats had been used multiple times in previous CO2 anesthesia studies, details not provided (42) SD, males, 1. CO2 17.25% VDR Euthanasia (CO2 Multiple, details not Control: each animal was No head shaking in any 400–500 g 2. Ar 17.25% VDR only) provided tested with air as a control animal 3. air 17.25% VDR Rearing, escape behaviors (n = 8) and vocalizations: increased with CO2 No LORR or death with Ar (43) Wistar, males, 1. CO2 17% VDR Gas exposure with seven times with 19 Control: each animal was Approach-avoidance: 400–500 g 2. CO2 20% recovery additional air/CO2 tested with air as a control number of cereal pieces 3. CO2 15% training trials eaten + latency to leave 4. CO2 10% decreased with increasing 5. CO2 5% [CO2 ], no rats ate cereal 6. Ar 90% (n = 9) and time dwelling only 3 s for Ar trials (44) Wistar, males, 1. CO2 17% VDR Gas exposure with >8 times Control: each animal was Rearing and nose to lid: ND 20–52 week (data used from recovery -rats had also been tested with air as a control Escape behaviors: previous trial) used multiple times increased from baseline with 2. peppermint in air in previous CO2 increasing [CO2 ] 17% VDR (n = 7–9) anesthesia studies, Approach-avoidance: details not provided number of cereal pieces eaten + latency to leave decreased with increasing [CO2 ] (45) Wistar, males, 32 1. CO2 27% VDR Gas exposure with At least 10 times Control: each animal was Approach-avoidance: week 2. CO2 20% VDR recovery -rats had also been tested with air as a control number of cereal pieces 3. CO2 14% VDR used multiple times eaten + latency to leave 4. CO2 7% VDR in previous CO2 decreased with increasing 5. CO2 3% VDR anesthesia studies, [CO2 ] (n = 8) details not provided (46) SD, males, 1. CO2 prefill Definitive: Rats exposed to Control: air only controls (n Rats exposed to Ar or N2 200–224 g 2. Ar prefill euthanasia three gasses on 3 = 9) but had been exposed had muscle 3. N2 prefill days for pilot study, to three gasses previously spasms/convulsions prior to 4. CO2 30% animals re-used in -rats instrumented for LOP 5. Air only (n = 5–7) definitive study telemetry under Heart rate: decreased for ketamine/xylazine CO2 groups with time anesthesia Blood pressure: decreased for CO2 groups with time LORR: fastest for CO2 prefill time to death: fastest for CO2 prefill (47) SD, males, 1. CO2 100% prefill Euthanasia None Rats instrumented with No vocalization or escape 271–391 g 2. CO2 gradual fill transducers for heart rate behaviors in either group (n = 4–5) and blood pressure under All rats in CO2 prefill group methohexitone urinated LOP: fastest for prefill CO2 Heart rate: decreased for CO2 groups with time Blood pressure: decreased for CO2 groups with time Time to death: fastest for CO2 prefill (Continued) Frontiers in Veterinary Science | www.frontiersin.org 10 July 2020 | Volume 7 | Article 411 Turner et al. CO2 Euthanasia of Laboratory Rodents TABLE 2 | Continued References Strain, Sex, and Intervention Euthanasia or Frequency of Control intervention Outcome measure Age treatments of anesthetic exposure or (group size) (identification of effect) interest (group induction with re-exposure size) recovery (48) SD, males, 8 week 1. CO2 24% FR Gas exposure with None Control: animals served as Aversion to light vs. dark: 2. isoflurane 5%/O2 recovery their own controls during more likely to stay in dark -Experiment 2 training trials compartment until only—at low vs. high recumbent with single light intensity levels exposure to iso vs. CO2 , (n = 8) more likely to leave isoflurane faster upon re-exposure ACE, acepromazine; ACTH, adrenocorticotropin hormone; cort, corticosterone; FR, flow rate; iso, isoflurane; LOP, time to loss of posture; LORR, time to loss of righting reflex; MDZ, midazolam; ND, no difference; pentobarb, pentobarbital; SD, Sprague-Dawley; VDR, volume displacement rate. TABLE 3 | Characteristics of the five neonatal rodent studies included in the systematic review. References Species, Strain, and Age Intervention of interest Outcome measures (identification of (group size) effect) (21) Mouse, various, strains, 1. CO2 20% VDR LORR: pentobarb faster than CO2 in both fetal pups embryonic day 2. Na pentobarb IP, 800 PND 1–7 and 8–14 pups 14–20 (see note), postnatal mg/kg (PND 8–14 only) Cardiac arrest: CO2 faster than pentobarb day (PND) 1–7, PND 8–14 3. Halothane 5%/O2 0.4 (evaluated only in PND 8–14 pups) Note: L/min (PND 1–7 n = 11, fetal mice (embryonic day 14–20) were PND 8–14 n = 15) also evaluated. Dams were euthanized with the three agents listed or via cervical dislocation with or without halothane anesthesia. For all methods, fetal pups were still alive at 20 min post-euthanasia/exposure of the dam. (49) Mice, various (including 1. Cervical dislocation of Dam treatments—time to fetus death: CD1, C57BL/6J, 129S2), dam pentobarb = CO2 , sodium pentobarbital ED 16 2. CO2 exposure of dam < cervical dislocation; no difference (details not specified) between dam strain 3. Na pentobarb IP, 120 Fetal treatment—time to fetal death: mg/kg to dam pentobarb intraplacental < pentobarb IP 4. Na pentobarb to dam = CO2 dam < cervical dislocation intraplacental, 600 mg/kg of dam; B6 = 129S2, B6 < CD1 (n = 9) (50) Mice, various strains, PND 1. Plastic bag pre-filled with Time to death: 60 min for pups PND 0–6, 0–10 100% CO2 (n = 5–10 per 20 min for PND 7–10, recovery of pups run, n = 2,355 total) after prolonged exposures common (51) Rat, SD and F344, PND 1. Plastic bag pre-filled with Time to death: 35 min for pups PND 0, 0–10 100% CO2 (n = 10 per run, 5 min for pups PND 10, recovery of pups n = 791 total) after prolonged exposures common (52) Mice, C57BL/6 and CD1, 1. Plastic bags placed LORR 2 min, exposed for 30 min and then PND 1–2 0.5 mL isoflurane-soaked following 30–120 min post-exposure cotton into bag and period 18/76 pups recovered resealed (n = 3–18, 76 total) IP, intraperitoneal; LORR, time to loss of righting reflex; pentobarb, pentobarbital; PND, postnatal day; VDR, volume displacement rate. between studies conducted within one laboratory (anesthesia- or papers included for review for mice, one evaluated exposure to exposure-only type studies). sublimation of dry ice (16), a technique no longer considered humane for rodent euthanasia because of the variability of CO2 gas production and the serious risk of freezing burns in Effects of Carbon Dioxide Inhalation on animals if they touch the surface of dry ice (−78.5◦ C) (1). Eight Adult Mice studies (53%) evaluated outcomes in mice following a single The study characteristics for mice were remarkably diverse, exposure trial in which euthanasia resulted, while six studies emphasizing that comparisons were difficult to make. Of the 15 (40%) evaluated outcomes following repeated exposures to CO2 Frontiers in Veterinary Science | www.frontiersin.org 11 July 2020 | Volume 7 | Article 411 Turner et al. CO2 Euthanasia of Laboratory Rodents FIGURE 2 | Risk of bias assessment summary for the 16 single trial studies in adult mice and rats, and rodent pups (16–19, 21, 25, 27, 28, 30, 32, 34, 35, 37, 39, 47, 49). FIGURE 3 | Quality assessment summary for the 37 papers included in this systematic review. (a) Reporting of any randomization, (b) reporting of any blinding, and (c) sample size calculation performed. or other inert or anesthetic gasses. When repeated exposures studies. One study (7%) incorporated untreated controls into occurred, they ranged from 3 to 50 or more exposures, sometimes the study design, four studies (27%) had no control comparator over a 4 month test period. Multiple strains of mice were group, and three studies (20%) used baseline values from the evaluated across the different studies (although generally only same animals as internal controls while seven (47%) induction- one per study) in males (47%), females (33%), or both (20%) sexes only studies used a variation of a cross-over designs in which mice of mice. Animals ranged in age from 8 weeks to over 24 weeks of were exposed to at least two or more gasses or inhalant agents, but age and studies sometimes were conducted opportunistically (2 not necessarily all tested agents. of 15 studies), using animals remaining after other experiments Studies evaluated, tested, and reported exposures to CO2 conducted by other research groups had been completed. The and other inhalant gasses in a variety of ways. Exposure to nature of the previous work with the animals was generally not CO2 or other inert or inhalant anesthetic gas was expressed specified nor was the interval or time period always defined as a percentage when delivered by face mask, and as volume between the previous studies and the euthanasia/anesthesia displacement rate (VDR; 40% of studies) and/or as flow rate (FR; Frontiers in Veterinary Science | www.frontiersin.org 12 July 2020 | Volume 7 | Article 411 Turner et al. CO2 Euthanasia of Laboratory Rodents 40% of studies) when delivered into a chamber or cage. Volume mice exposed to CO2 (20–50% FR) prior to LOP (25), but was displacement rate is considered the most accurate exposure not reported in a subsequent study by the same researchers using method for calculating CO2 delivery into a chamber because of 20% VDR CO2 or isoflurane exposure (27). the marked variability in cage and chamber size (1). Because In studies evaluating physiologic parameters in mice, time chamber dimensions were not routinely provided FR could to LORR following CO2 exposure were consistently in three not be converted to VDR. Similarly, gas flow patterns within circumstances: when compared to other tested agents, such as the chamber will markedly affect the outcome following gas isoflurane in oxygen (18, 27), for increasing VDRs or flow rates exposure, e.g., use of a gas diffuser and the specific location of CO2 (17, 28, 30), and when CO2 was combined with nitrous of gas introduction into the chamber, since all gasses studied oxide (N2 O) (29). Use of a sedative, such as acepromazine or are heavier than air. Some studies evaluated different VDRs or midazolam, prior to administering CO2 did not shorten time to FRs of CO2 , some compared one or more VDR/FR of CO2 LORR (30). Heart rates and mean blood pressure increased from with one or more inert or inhalant anesthetic gas and some baseline prior to the onset of ataxia or LORR in all mice exposed studies compared one or more VDR/FR/percentage of CO2 with to VDRs of CO2 between 15 and 100% and for mice exposed to other physical or chemical methods of euthanasia, including isoflurane in oxygen (17, 18). There was no difference in peak pentobarbital sodium, potassium chloride, cervical dislocation, heart rate or mean blood pressure with different VDRs of CO2 and decapitation. One study evaluated the effects of different (15–100%), although, time to peak cardiovascular parameters VDRs of CO2 with or without premedication with acepromazine was shorter with increasing VDRs of CO2 (17). Peak heart or midazolam (30) and one study evaluated the effects of different rates were higher for mice exposed to isoflurane compared with flow rates of CO2 when given with differing ambient light different VDRs of CO2 , although peak mean blood pressure intensities (28). Flow or volume displacement rates of carbon did not differ for mice exposed to CO2 vs. isoflurane (18). dioxide evaluated ranged from 0.2 L/min to 15–100%. No differences were noted in plasma ACTH levels between A wide range of animal welfare-related outcome measures mice exposed to different VDRs of CO2 vs. isoflurane (17, 18) were reported, and the outcomes and direction of effect are listed and increased plasma corticosterone levels were noted for mice in Table 1. When determining welfare impacts of exposure to exposed to isoflurane for induction vs. CO2 alone (28). Time to CO2 or another method, the most important considerations are loss of EEG signal and loss of visual evoked potential was fastest time to loss of consciousness and signs or changes suggestive for mice exposed to 100% CO2 by facemask (vs. 70% CO2 ) (19). of pain, discomfort or distress that are seen prior to loss of Five studies did not examine time to loss of consciousness, consciousness, since anything that occurs after this time will anesthesia or euthanasia following exposure to CO2 or other not be perceived by the animal. Assessment of unconsciousness inert or inhalant gasses, but instead evaluated effects of repeated was not conducted in the same way in studies that assessed exposure to different VDRs or FRs of CO2 in approach-avoidance this parameter. Loss of posture (LOP) or “nose down,” and loss studies. These results were sometimes compared to argon (alone of righting reflex (LORR) were all used by different research or in combination with CO2 ), carbon monoxide (CO), isoflurane, groups. At high flow rates, LOP occurs almost simultaneously and/or sevoflurane (20, 22–24, 26). VDRs, FRs, and percentage with LORR; however, LOP can occur 5–20 s prior to LORR inhalant gas and rates of carrier oxygen were significantly during induction for anesthesia. Assessment of LORR requires different between these studies making comparisons difficult. an experimental set-up that permits the observer to handle and In one study evaluating the impact of CO2 exposure (53% FR) manipulate the animal. compared to argon (99% FR) and mixtures of CO2 with argon, Of the 20 outcome measures evaluated in mice across the 15 the time dwelling and time to withdrawal from chambers in studies, 10 were behavioral and included one or more of chamber which mice were exposed to gasses were shorter for mice exposed escape attempts, “anxiety/pain” behaviors, jumping, increased to CO2 alone or in combination with argon (22). In a subsequent activity, distance traveled, vocalization, rearing, urination, study by this group comparing different FRs of CO2 (25.5– defecation, and labored breathing. One outcome measure was 50.8% FR) alone to argon (99% FR), there was no difference a trained response requiring repeated exposures to CO2 and in the chamber dwelling time and time to withdrawal from the other agents over training and definitive trials, that is, using chamber between any of the treatment groups (23). Guedes et al. approach-avoidance techniques. The remaining nine outcomes (20) compared pairs of gasses (sevoflurane to CO2 , sevoflurane were physiologic in nature and included one or more of time to to isoflurane, and isoflurane to CO2 ) and found no difference loss of posture or nose down, time to loss of righting reflex, heart in dwelling time in the gas-only chambers between any of the rate, mean blood pressure, plasma corticosterone level, plasma pairs of gasses. When time to withdrawal from the chamber ACTH level, time to loss of EEG signal, time to loss of visual for CO2 (20% VDR) exposure was compared to isoflurane in evoked potential, and time to death. oxygen vs. isoflurane drops on a cotton ball, withdrawal times In terms of behavioral findings, mice exposed to isoflurane were longest for isoflurane in oxygen and shorter for CO2 and demonstrated more movement in the chamber during induction isoflurane drops (26). However, upon re-exposure to the same than mice exposed to different VDRs of CO2 , but there was agents, time to withdrawal from the chamber was shortest for no difference in escape attempts, pain or anxiety behaviors, isoflurane in oxygen (26). Finally, when mice were exposed 50 or including urination, defecation, and rearing, or ultrasonic more times to different gasses, including 70% FR CO2 vs. argon vocalizations by mice induced with either CO2 or isoflurane (16– vs. CO vs. isoflurane, latency to withdraw from a chamber in 18, 22, 23, 28–30). In one study, labored breathing was noted in which they received food rewards always occurred when CO2 Frontiers in Veterinary Science | www.frontiersin.org 13 July 2020 | Volume 7 | Article 411 Turner et al. CO2 Euthanasia of Laboratory Rodents chamber concentrations reached 13.5–18.2%, latency time to of posture (LOP) and loss of righting reflex (LORR) were used in leave was equal for isoflurane and CO, and latencies were longest different studies. for argon (24). Of the 21 outcome measures evaluated in rats across the 21 studies, 10 were behavioral and included at least one of: chamber escape attempts, jumping/rearing, increased activity, distance traveled, vocalization, urination, defecation, Effects of Carbon Dioxide Inhalation on labored breathing/gasping/head shaking, nasal bleeding, and Adult Rats seizures. One of the outcome measures assessed was a trained Similar to what was noted for mice, the study characteristics response requiring repeated exposures to CO2 and other agents were highly diverse for rats. Of the 21 papers included for over training and definitive trials using approach-avoidance review for rats, one evaluated exposure to sublimation of dry ice techniques. The remaining 10 outcomes were physiologic in (16), a procedure no longer considered acceptable, as mentioned nature and included at least one of: time to LOP, time to LORR, above. This paper was not considered further. Ten studies heart rate, mean blood pressure, plasma corticosterone level, (48%) evaluated outcomes in rats following a single definitive plasma ACTH level, serum glucose levels, time to isoelectric EEG exposure trial in which euthanasia resulted (although in 3 of signal, changes in electromyograph signal, and time to death. these studies, pilot studies involving gas exposures were also For behavioral outcomes, several studies reported no adverse conducted using the same study animals), while another 10 outcomes in rats (32, 38, 47), while other researchers reported studies (48%) evaluated outcomes following repeated exposures nasal bleeding (36) or gasping (35, 39) at very high exposures of to CO2 or other inert or anesthetic gasses. When repeated CO2 (60% FR or greater). Use of acepromazine as a sedative prior exposures occurred, the number of prior exposures was often to administering CO2 did not provide any apparent extra benefit, not specified. Eight of the studies (38% of the total) (31, 40– as these researchers did not observe any urination, defecation, 45, 48) were conducted by one lab and, as mentioned, many vocalization in rats exposed to CO2 at 6 L/min FR (37). Similarly, of the same rats were re-used in multiple studies with results in the same study, tachypnea was seen in rats with increasing published in separate papers. Studies were conducted largely in chamber concentrations of CO2 , regardless of whether rats had Wistar or Sprague-Dawley (SD) rats and in both sexes, although been premedicated with acepromazine as a sedative (37). generally only one sex per study (71% were conducted in males In euthanasia studies evaluating physiologic responses, time only, 19% in females only, 5% in both sexes, and 5% did not to LORR, times were shortest for argon when supplied at 50% specify animal sex). In these studies, animals ranged in age VDR or higher (32); however, seizures were reported in rats from 7 to 52 weeks. Similar to mice, details concerning previous prior to loss of consciousness when very high exposures to argon experimental work with the rats was poorly specified and animals (or nitrogen) occurred (46). Otherwise, LORR was fastest when were used opportunistically in at least four studies. None of rats were placed in novel chambers prefilled with 100% CO2 the rat studies incorporated untreated controls into the study (35, 36, 38, 39, 46, 47). Only one of the euthanasia studies design, four studies (19%) had no control comparator group, and compared the effects of CO2 to isoflurane induction, followed 15 studies (71%) used baseline values from the same animals by CO2 (34), and inductions (LORR) were noted to be slower as internal controls and/or used the same animals in pilot with isoflurane compared with 20% CO2 VDR. Heart rates and trials. In seven of the euthanasia studies (33%), rats had been mean blood pressure decreased from baseline in all rats exposed previously surgically instrumented with catheters or telemetry to VDRs of CO2 between 10 and 100% and for rats exposed transducers under anesthesia (which included ketamine/xylazine, to isoflurane in oxygen (32, 34, 35, 39, 46). In all studies in methohexitone, isoflurane, or sodium pentobarbital). Only one which EEGs were evaluated, time to isoelectric EEG and time to of these studies (34) re-exposed rats to the same anesthetic agent death were shortest for exposures to CO2 of 20% VDR or greater (isoflurane) during the definitive euthanasia trials. (34, 35, 39, 46, 47). In one study, exposure to argon at 50% VDR Similar to studies in mice, a range of methods was used was compared to exposure to CO2 at 10% VDR, and time to for reporting exposures to CO2 and other gasses. Some studies death was faster for rats exposed to argon (32). No differences evaluated prefilled chambers of CO2 vs. different VDRs or FRs of were noted in serum glucose in rats exposed to CO2 at 6 L/min CO2 , some compared one or more VDR/FR of CO2 (sometimes when compared with rats exposed to the same level of CO2 but mixed with oxygen or nitrogen) compared to one or more inert pre-treated with either acepromazine or pentobarbital; however, (e.g., argon, nitrogen) or inhalant anesthetic gas (isoflurane, the rats exposed first to pentobarbital had higher serum levels of sevoflurane), and one study compared one or more VDR or ACTH and corticosterone (37). FR of CO2 with pentobarbital sodium euthanasia (34). One Ten studies reviewed did not examine time to loss of study evaluated the effects of different VDRs of CO2 with or consciousness, anesthesia or euthanasia following exposure to without premedication with acepromazine (37). Flow or volume CO2 or other inert or inhalant gasses, but instead evaluated displacement rates of carbon dioxide evaluated ranged from 3% effects of repeated exposure to different VDRs or FRs of CO2 or VDR to 100% CO2 chamber prefill. other inert or anesthetic gasses in approach-avoidance studies. A wide range of animal welfare-related outcome measures These results were sometimes compared to argon (alone or in were reported, and the outcomes and direction of effect are listed combination with CO2 and/or oxygen) or isoflurane and/or in Table 2. For rat studies, assessment of unconsciousness was sevoflurane (22, 23, 31, 40, 41, 43–46, 48). VDRs, FRs, and not conducted in the same way between studies, and both loss percentage inhalant gas and rates of carrier oxygen were also Frontiers in Veterinary Science | www.frontiersin.org 14 July 2020 | Volume 7 | Article 411 Turner et al. CO2 Euthanasia of Laboratory Rodents significantly different between these studies making comparisons 30–120 min of room air exposure (52). Rat pups also required difficult. Similar to mice, in one study evaluating the impact prolonged exposures to CO2 gas for euthanasia, although rats of CO2 exposure (53% FR) compared to argon (99% FR) and succumbed more quickly at PND 0 (35 min) than mice (60 min) mixtures of CO2 with argon, the dwelling time in the chamber (51). In summary, these findings indicate that very prolonged and time to withdrawal from chamber in which rats were exposed exposures are required when CO2 exposure is used for euthanasia to gasses were shorter for rats exposed to CO2 alone or in of mouse or rat pups that are PND 6 or younger. combination with argon (22). In a subsequent study by this group comparing different FRs of CO2 (25.5–50.8% FR) alone to argon DISCUSSION (99% FR), there was no difference in the dwelling time and time to withdrawal between any of the treatment groups (23). When Death is a critical and permanent juncture in the course of different flow rates of argon alone were compared, chamber any animal’s life. Therefore, there is significant interest in dwelling time (i.e., latency to leave) decreased with increasing ensuring that methods used for euthanasia of laboratory mice and concentrations of argon (41). When aversion-avoidance was rats are humane. Although many organizations have developed evaluated for isoflurane compared to sevoflurane, both agents guidelines and recommendations for euthanasia of laboratory were determined to be aversive and aversion increased with rodents [for example, (1, 53)], to date, these guidelines have re-exposure (31). Further, when different CO2 VDRs or FRs not been based on a systematic review of the evidence available. were compared in an approach-avoidance paradigm, chamber This systematic review is the first to examine the evidence dwelling times (i.e., latency to leave) were reduced as CO2 for the welfare impact of CO2 inhalation for euthanasia of concentrations increased (40, 43, 45) or when compared to laboratory mice and rats alone or in comparison with other exposure to isoflurane (48). euthanasia methods. Available Evidence and Quality Effects of Carbon Dioxide Inhalation on Studies evaluating the behavioral and physiologic effects of CO2 Rodent Pups gas inhalation for euthanasia of laboratory mice, rats, and rodent When evaluating the effects of carbon dioxide inhalation on pups were highly heterogenous in approach. Heterogeneity was rodent pups, a variety of study designs and comparators caused by differences in the populations (sex of animal, different were used. Five studies examined the effect of CO2 gas stocks and strains, prior use of animals in previous experiments, administration, isoflurane exposure, cervical dislocation, or ages of adult animals), the interventions studied (CO2 gas; sodium pentobarbital administered by intraperitoneal (IP) other anesthetic or inert gasses; physical methods; chemical injection to the dam or by intraplacental injection (fetal mice methods; rate, method, and reporting of gas introduction and only) on neonatal and fetal rodent pups. Four of the studies maintenance; age of perinatal rodent pups), and the outcome evaluated euthanasia of mouse pups (neonatal or embryonic) measures evaluated within and between studies. Most of the (21, 49, 50, 52) and one study evaluated the effects of CO2 reported outcomes for single trial euthanasia studies in laboratory gas exposure on neonatal rat pups (51). One study discussed mice and rats showed no adverse effect of CO2 on behavioral the prolonged time to death of neonatal pups after exposure to or physiologic outcomes when administered to animals in a halothane, a halogenated anesthetic no longer available in many chamber at 70% VDR or less, and responses were generally no countries (21). Multiple stocks and strains of mice were evaluated different from baseline. Evidence indicating discomfort prior to by the different neonatal mouse studies and SD stock and F344 induction for euthanasia is present for CO2 gas when supplied strain were evaluated in the neonatal rat study. to mice or rats in 100% prefilled chambers as gasping or The welfare outcomes measured for the rodent pups or fetuses tachypnea. Distress ultrasonic vocalizations (26.5 kHz) were also for these studies included time to loss of righting reflex (LORR; noted in mice during single trial euthanasia studies when exposed pups only), time to cardiac arrest, and time to death (Table 3). to CO2 in 100% prefilled chambers but not at lower VDRs. Administration of sodium pentobarbital IP to postnatal day Similar distress ultrasonic vocalizations were noted when mice (PND) 8–14 mouse pups resulted in a faster LORR than for were induced with isoflurane in oxygen. When exposed to exposure of this same age group to CO2 gas; however, time subanesthetic concentrations of CO2 gas over multiple training to cardiac arrest was faster for CO2 gas exposure (21). Fetal trials, mice and rats demonstrated avoidance behaviors in mouse death was fastest following intraplacental administration test trials. Avoidance behaviors were also seen in mice and of sodium pentobarbital compared to either administration of rats exposed to subanesthetic concentrations of isoflurane and sodium pentobarbital IP to dams or exposure of dams to CO2 gas sevoflurane with multiple training trials. Reliability and overall (49). Very young mouse pups (PND 0–6) of both sexes required interpretation of these findings is hampered by a number of prolonged exposures to CO2 to induce death (up to 60 min for limitations. In the one study in mice comparing CO2 euthanasia PND 0)—if shorter times were used it was common for pups to (100% prefill) to physical methods, the time to loss of EEG signal recover when exposed to room air (50). Further, inbred strains was similar between CO2 , KCl injection, cervical dislocation, were more resistant to the lethal effects of CO2 gas inhalation and decapitation. than outbred stocks (50). Exposure to high concentrations of Not all of the studies in the body of literature that isoflurane gas in oxygen for up to 30 min resulted in anesthesia are frequently cited as evidence of CO2 aversion evaluated of PND 1–2 mouse pups, but 24% of pups recovered after physiologic or behavioral variables preceding euthanasia but Frontiers in Veterinary Science | www.frontiersin.org 15 July 2020 | Volume 7 | Article 411 Turner et al. CO2 Euthanasia of Laboratory Rodents rather evaluated parameters in response to exposures to apparent aversion/avoidance between these agents and CO2 are subanesthetic CO2 or other gasses. Some of the studies reviewed not always clear cut. For example, repeated exposure to isoflurane evaluated behaviors that required repeated exposures to gas was evaluated as being more aversive than repeated exposure to for reliable performance of a given behavior, and mice and CO2 gas (26, 48, 60). rats in many studies were re-used between pilot and definitive Further studies are needed to accurately assess the impact phases, between different trials within the same study, and in of inhalant euthanasia methods, such as exposure to CO2 gas some cases, between different experiments conducted within on laboratory mice and mouse welfare, including the impact of the same lab and published in separate papers. These repeat euthanizing mice and rats in their home cages. A strategy for exposure experimental designs were likely used in an attempt further research in this area has recently been published (61). to reduce the numbers of animals needed, an important 3Rs consideration. While repeat exposure studies can inform the CONCLUSION research community about aspects of animal welfare, they cannot be considered as strictly equivalent to single trial There is insufficient evidence to permit an unbiased assessment euthanasia studies. of the overall impact of CO2 inhalation during euthanasia on Appropriate reporting of methods is essential for interpreting welfare indicators in laboratory mice and rats. Additional well- study bias and quality in all biomedical research publications designed, unbiased, and adequately powered studies are needed and to ensure reproducibility (54, 55). This has been emphasized to accurately assess the welfare impact of CO2 gas euthanasia repeatedly to the research community with publication of method for laboratory mice and rats and to identify alternative guidelines to support conduct and reporting of high quality techniques that represent a significant improvement or benefit to experiments (56, 57). Our risk of bias assessment indicated animal welfare. generally poor or unclear bias in most of the studies for which this tool could be used. Furthermore, the overall study quality DATA AVAILABILITY STATEMENT assessment suggested generally poor quality of evidence for the welfare effect of CO2 and other inhaled gasses for euthanasia All datasets presented in this study are included in the of laboratory mice and rats, when considering sample size article/Supplementary Material. determination, and risks of performance and detection bias. Sample size calculations force research teams to define a primary AUTHOR CONTRIBUTIONS outcome for their study a priori and the effect size needed to detect a difference between groups. When this is missing or All authors contributed to conception and design of the study. poorly done, it becomes difficult to interpret the relevance of PT, JL, and MR-H developed the SR protocol and submitted effects noted. For example, a statistically significant result within to SYRCLE for approval. PT, DH, and TK conducted the a small group of animals (e.g., a 5–10 s difference in loss of literature screening. PT and JS developed the data extraction righting reflex between groups) may not be biologically relevant questionnaire. PT and DH conducted the data extraction and when the entire population is considered (58). Performance and ROB analysis. PT wrote the first draft of the manuscript. detection bias are particularly critical for euthanasia procedures, All authors contributed to manuscript revision, and read and in that euthanasia is a procedure that many people find distasteful approved the submitted version. and challenging to perform. Recently, a clear demonstration of bias occurred when study participants were asked to score the FUNDING quality of observed loss of consciousness of mice and rats. The scoring differed when viewers were told that animals were being Publication of this paper was supported by IACLAM. euthanized or anesthetized. (59). This demonstrates that it is essential that appropriate blinding occur when collecting and then assessing euthanasia outcomes. ACKNOWLEDGMENTS Because of these general shortcomings in study design and We would like to thank Deirdre Stuart and Brianna Mercer, conduct, the results of this systematic review indicate that there summer research assistants at the University of Guelph, for is insufficient evidence to permit an unbiased assessment of technical support with the literature review process. the impact of CO2 inhalation during euthanasia on welfare indicators in laboratory mice and rats. Further, while studies of repeated exposures to CO2 gas suggest that CO2 induction SUPPLEMENTARY MATERIAL using prefilled containers may result in short periods of distress, induction with accepted inhalant anesthetic agents, including The Supplementary Material for this article can be found isoflurane and sevoflurane in oxygen, also are aversive to online at: https://www.frontiersin.org/articles/10.3389/fvets. laboratory mice and rats. The aversion/avoidance differences 2020.00411/full#supplementary-material between treatment groups are small under highly controlled Supplementary Table 1 | PRISMA Checklist. laboratory conditions, and positive vs. negative differences in Supplementary Table 2 | Comprehensive Search Strategy. Frontiers in Veterinary Science | www.frontiersin.org 16 July 2020 | Volume 7 | Article 411 Turner et al. CO2 Euthanasia of Laboratory Rodents REFERENCES 24. Makowska IJ, Vickers L, Mancell J, Weary DM. Evaluating methods of gas euthanasia for laboratory mice. Appl Anim Behav Sci. (2009) 121:230– 1. AVMA. AVMA Guidelines for the Euthanasia of Animals. 2020 ed. American 5. doi: 10.1016/j.applanim.2009.10.001 Veterinary Medical Association (2020). Available online at: https://www. 25. Moody CM, Chua B, Weary DM. The effect of carbon dioxide flow avma.org/kb/policies/documents/euthanasia.pdf (accessed January 25, 2020). rate on the euthanasia of laboratory mice. Lab Anim. (2014) 48:298– 2. Boivin GP, Hickman DL, Creamer-Hente MA, Pritchett-Corning KR, 304. doi: 10.1177/0023677214546509 Bratcher NA. Review of CO2 as a euthanasia agent for laboratory rats and 26. Moody CM, Weary DM. Mouse aversion to isoflurane versus mice. J Am Assoc Lab Anim Sci. (2017) 56:491–9. carbon dioxide gas. Appl Anim Behav Sci. (2014) 158:95– 3. Merriman JE. The role of carbon dioxide in anaesthesia. Can Anaesth Soc J. 101. doi: 10.1016/j.applanim.2014.04.011 (1995) 2:273–80. doi: 10.1007/BF03016171 27. Moody CM, Makowska IJ, Weary DM. Testing three measures of mouse 4. Wilson SR. Carbon dioxide in anaesthesia. Br J Anaesth. (1926) 3:112– insensibility following induction with isoflurane or carbon dioxide gas 8. doi: 10.1093/bja/3.3.112 for a more humane euthanasia. Appl Anim Behav Sci. (2015) 163:183– 5. Bert P. La Pression Barometrique. Paris: G. Masson (1978). 610 p. 7. doi: 10.1016/j.applanim.2014.11.010 6. Permentier K, Vercammen S, Soetaert S, Schellemans C. Carbon 28. Powell K, Ethun K, Taylor DK. The effect of light level, CO2 flow rate, and dioxide poisoning: a literature review of an often forgotten cause of anesthesia on the stress response of mice during CO2 euthanasia. Lab Anim. intoxication in the emergency department. Int J Emerg Med. (2017) (2016) 45:386–95. doi: 10.1038/laban.1117 10:14. doi: 10.1186/s12245-017-0142-y 29. Thomas AA, Flecknell PA, Golledge HD. Combining nitrous oxide 7. Langford NJ. Carbon dioxide poisoning. Toxicol Rev. (2005) 24:229– with carbon dioxide decreases the time to loss of consciousness during 35. doi: 10.2165/00139709-200524040-00003 euthanasia in mice - refinement of animal welfare? PLoS One. (2012) 8. Conlee KM, Stephens ML, Rowan AN, King LA. Carbon dioxide for 7:e32290. doi: 10.1371/journal.pone.0032290 euthanasia: Concerns regarding pain and distress, with special reference to 30. Valentine H, Williams WO, Maurer KJ. Sedation or inhalant anesthesia before mice and rats. Lab Anim. (2005) 39:137–61. doi: 10.1258/0023677053739747 euthanasia with CO2 does not reduce behavioral or physiologic signs of pain 9. Hawkins P, Prescott MJ, Carbone L, Dennison N, Johnson C, Makowska IJ, and stress in mice. J Am Assoc Lab Anim Sci. (2012) 51:50–7. et al. A good death? Report of the Second Newcastle Meeting on laboratory 31. Boulanger-Bertolus J, Nemeth G, Makowska IJ, Weary DM. Rat aversion animal euthanasia. Animals (Basel). (2016) 6:50. doi: 10.3390/ani6090050 to sevoflurane and isoflurane. Appl Anim Behav Sci. (2014) 164:73– 10. Axiak Flammer S, Eskes C, Kohler I, Ochiang Pernet A, Jakob P, 80. doi: 10.1016/j.applanim.2014.12.013 Marahrens M, et al. Alternatives to carbon dioxide - taking responsibility 32. Burkholder TH, Niel L, Weed JL, Brinster LR, Bacher JD, Foltz CJ. for humanely ending the life of animals. Animals (Basel). (2019) Comparison of carbon dioxide and argon euthanasia: effects on behavior, 9:482. doi: 10.3390/ani9080482 heart rate, and respiratory lesions in rats. J Am Assoc Lab Anim Sci. 11. Ziemann AE, Allen JE, Dahdaleh NS, Drebot II, Coryell M, Wunsch AM, et al. (2010) 49:448–53. The amygdala is a chemosensor that detects carbon dioxide and acidosis to 33. Chisholm J, De Rantere D, Fernandez NJ, Krajacic A, Pang DS. Carbon elicit fear behavior. Cell. (2009) 139:1012–21. doi: 10.1016/j.cell.2009.10.029 dioxide, but not isoflurane, elicits ultrasonic vocalizations in female rats. Lab 12. National Research Council. Guide for the Care and Use of Laboratory Animals. Anim. (2013) 47:324–7. doi: 10.1177/0023677213493410 8th ed. Washington, DC: The National Academies Press (2011). 105 p. 34. Chisholm JM, Pang DS. Assessment of carbon dioxide, carbon 13. Moher D, Liberati A, Tetzlaff J, Altman DG, the PRISMA Group. Preferred dioxide/oxygen, isoflurane and pentobarbital killing methods reporting items for systematic reviews and metaAnalyses: the PRISMA in adult female Sprague-Dawley rats. PLoS One. (2016) statement. PLoS Med. (2009) 6:e1000097. doi: 10.1371/journal.pmed.1000097 11:e0162639. doi: 10.1371/journal.pone.0162639 14. Turner PV, Sargeant J, Hickman D, Kurosawa TM, Mercer B, Ritskes M, 35. Coenen AM, Drinkenburg WH, Hoenderken R, van Luijtelaar et al. Protocol for Systematic Review of Carbon Dioxide in Laboratory Rodents. EL. Carbon dioxide euthanasia in rats: oxygen supplementation The Netherlands: SYRCLE (2017). Available online at: https://issuu.com/ minimizes signs of agitation and asphyxia. Lab Anim. (1995) radboudumc/docs/the_use_of_carbon_dioxide_as_a_meth (accessed January 29:262–8. doi: 10.1258/002367795781088289 25, 2020). 36. Danneman PJ, Stein S, Walshaw SO. Humane and practical implications of 15. Hooijmans CR, Rovers MM, de Vries RB, Leenaars M, Ritskes-Hoitinga M, using carbon dioxide mixed with oxygen for anesthesia or euthanasia of rats. Langendam MW. SYRCLE’s risk of bias tool for animal studies. BMC Med Res Lab Anim Sci. (1997) 47:376–85. Methodol. (2014) 14:43. doi: 10.1186/1471-2288-14-43 37. Hackbarth H, Küppers N, Bohnet W. Euthanasia of rats with 16. Blackshaw JK, Fenwick DC, Beattie AW, Allan DJ. The behaviour of chickens, carbon dioxide - animal welfare aspects. Lab Anim. (2000) mice and rats during euthanasia with chloroform, carbon dioxide and ether. 34:91–6. doi: 10.1258/002367700780578055 Lab Anim. (1988) 22:67–75. doi: 10.1258/002367788780746674 38. Hewett TA, Kovacs MS, Artwohl JE, Bennett BT. A comparison of euthanasia 17. Boivin GP, Bottomley MA, Dudley ES, Schiml PA, Wyatt CN, Grobe N. methods in rats, using carbon dioxide in prefilled and fixed flow rate filled Physiological, behavioral, and histological responses of male C57BL/6N mice chambers. Lab Anim Sci. (1993) 43:579–82. to different CO2 chamber replacement rates. J Am Assoc Lab Anim Sci. 39. Hickman DL, Fitz SD, Bernabe CS, Caliman IF, Haulcomb MM, Federici (2016) 55:451–61. LM, et al. Evaluation of low versus high volume per minute displacement 18. Boivin GP, Bottomley MA, Schiml PA, Goss L, Grobe N. Physiologic, CO2 methods of euthanasia in the induction and duration of panic-associated behavioral, and histologic responses to various euthanasia methods in behavior and physiology. Anim. (2016) 6:E45. doi: 10.3390/ani6080045 C57BL/6NTac male mice. J Am Assoc Lab Anim Sci. (2017) 56:69–78. 40. Kirkden RD, Niel L, Stewart SA, Weary DM. Gas killing of rats: the effect of 19. Cartner SC, Barlow SC, Ness TJ. Loss of cortical function in mice after supplemental oxygen on aversion to carbon dioxide. Anim Welf. (2008) 17:79– decapitation, cervical dislocation, potassium chloride injection, and CO2 87. inhalation. Comp Med. (2007) 57:570–3. 41. Makowska IJ, Niel L, Kirkden RD, Weary DM. Rats show 20. Guedes SR, Valentim AM, Antunes LM. Mice aversion to sevoflurane, aversion to argon-induced hypoxia. Appl Anim Behav Sci. (2008) isoflurane and carbon dioxide using an approach-avoidance task. Appl Anim 114:572–81. doi: 10.1016/j.applanim.2008.04.005 Behav Sci. (2017) 189:91–7. doi: 10.1016/j.applanim.2017.01.012 42. Niel L, Weary DM. Behavioural responses of rats to gradual-fill carbon dioxide 21. Klaunberg BA, O’Malley J, Clark T, Davis JA. Euthanasia of mouse fetuses and euthanasia and reduced oxygen concentrations. Appl Anim Behav Sci. (2006) neonates. Contemp Top Lab Anim Sci. (2004) 43:29–34. 100:295–308. doi: 10.1016/j.applanim.2005.12.001 22. Leach MC, Bowell VA, Allan TF, Morton DB. Aversion to gaseous euthanasia 43. Niel L, Weary DM. Rats avoid exposure to carbon dioxide and argon. Appl agents in rats and mice. Comp Med. (2002) 52:249–57. Anim Behav Sci. (2007) 107:100–9. doi: 10.1016/j.applanim.2006.08.002 23. Leach MC, Bowell VA, Allan TF, Morton DB. Measurement of aversion 44. Niel L, Kirkden RD, Weary DM. Effects of novelty on rats’ to determine humane methods of anaesthesia and euthanasia. Anim Welf. responses to CO2 exposure. Appl Anim Behav Sci. (2008) (2004) 13:S77–86. 111:183–94. doi: 10.1016/j.applanim.2007.06.004 Frontiers in Veterinary Science | www.frontiersin.org 17 July 2020 | Volume 7 | Article 411 Turner et al. CO2 Euthanasia of Laboratory Rodents 45. Niel L, Stewart SA, Weary DM. Effect of flow rate on aversion to gradual- 57. Percie du Sert N, Ahluwalia A, Alam S, Avey MT, Baker M, Browne WJ, fill carbon dioxide exposure in rats. Appl Anim Behav Sci. (2008) 109:77– et al. The ARRIVE Guidelines 2019: Updated Guidelines for Reporting Animal 84. doi: 10.1016/j.applanim.2007.02.004 Research. (2019). Available online at: https://www.biorxiv.org/content/10. 46. Sharp J, Azar T, Lawson D. Comparison of carbon dioxide, argon, and 1101/703181v1 (accessed February 11, 2020). nitrogen for inducing unconsciousness or euthanasia of rats. J Am Assoc Lab 58. Faraone SV. Interpreting estimates of treatment effects. Implications for Anim Sci. (2006) 45:21–5. managed care. Pharm Therapeut. (2008) 33:700–11. 47. Smith W, Harrap SB. Behavioural and cardiovascular responses of rats 59. Baker BA, Hickman DL. Bias in rating of rodent distress during anesthesia to euthanasia using carbon dioxide gas. Lab Anim. (1997) 31:337– induction for anesthesia compared with euthanasia. J Am Assoc Lab Anim Sci. 46. doi: 10.1258/002367797780596130 (2018) 57:143–56. 48. Wong D, Makowska IJ, Weary DM. Rat aversion to isoflurane versus carbon 60. Marquardt N, Feja M, Hünigen H, Plendl J, Menken L, Fink dioxide. Biol Lett. (2013) 9:20121000. doi: 10.1098/rsbl.2012.1000 H, et al. Euthanasia of laboratory mice: are isoflurane and 49. Muñoz-Mediavilla C, Cámara JA, Salazar S, Segui B, Sanguino D, Mulero sevoflurane real alternatives to carbon dioxide? PLoS ONE. (2018) F, et al. Evaluation of the foetal time to death in mice after application 13:e0203793. doi: 10.1371/journal.pone.0203793 of direct and indirect euthanasia methods. Lab Anim. (2016) 50:100– 61. Steiner AR, Flammer SA, Beausoleil NJ, Berg C, Bettschart-Wolfensberger 7. doi: 10.1177/0023677215600626 R, Pinillos RG, et al. Humanely ending the life of animals: Research 50. Pritchett K, Corrow D, Stockwell J, Smith A. Euthanasia of neonatal mice with priorities to identify alternatives to carbon dioxide. Anim. (2019) carbon dioxide. Comp Med. (2005) 55:275–81. 9:E911. doi: 10.3390/ani9110911 51. Pritchett-Corning KR. Euthanasia of neonatal rats with carbon dioxide. J Am Assoc Lab Anim Sci. (2009) 48:23–7. Conflict of Interest: The authors declare that the research was conducted in the 52. Seymour TL, Nagamine CM. Evaluation of isoflurane overdose for euthanasia absence of any commercial or financial relationships that could be construed as a of neonatal rats. J Am Assoc Lab Anim Sci. (2016) 55:321–3. potential conflict of interest. 53. Canadian Council on Animal Care. Guidelines on Euthanasia of Animals Used in Science. (2010). Available online at: https://www.ccac.ca/Documents/ The reviewer MRL declared a past collaboration with the author DH to the Standards/Guidelines/Euthanasia.pdf (accessed February 11, 2020). handling editor. 54. Macleod MR, Lawson McLean A, Kyriakopoulou A, Serghiou S, de Wilde A, Sherratt N, et al. Risk of bias in reports of Copyright © 2020 Turner, Hickman, van Luijk, Ritskes-Hoitinga, Sargeant, in vivo research: a focus for improvement. PLoS Biol. (2015) Kurosawa, Agui, Baumans, Choi, Choi, Flecknell, Lee, Otaegui, Pritchett-Corning 13:e1002273. doi: 10.1371/journal.pbio.1002273 and Shimada. This is an open-access article distributed under the terms of 55. Reichlin TS, Vogt L, Würbel H. The researchers’ view of scientific rigor - the Creative Commons Attribution License (CC BY). The use, distribution survey on the conduct and reporting of in vivo research. PLoS One. (2016) or reproduction in other forums is permitted, provided the original author(s) 11:e0165999. doi: 10.1371/journal.pone.0165999 and the copyright owner(s) are credited and that the original publication in 56. Kilkenny C, Browne WJ, Cuthill IC, Emerson M, Altman DG. Improving this journal is cited, in accordance with accepted academic practice. No use, bioscience research reporting: the ARRIVE guidelines for reporting animal distribution or reproduction is permitted which does not comply with these research. PLoS Biol. (2010) 8:e1000412. doi: 10.1371/journal.pbio.1000412 terms. Frontiers in Veterinary Science | www.frontiersin.org 18 July 2020 | Volume 7 | Article 411