A variety of methods have been used to study ventilatory depression in humans, including Pao2 and Paco2 serial blood gas analysis (SBGA), pulse oximeter and capnometer monitoring (1), end-tidal carbon dioxide (CO2) measurements with ventilatory and occlusion pressure responses to CO2 re-breathing (2), and continuous respiratory inductance plethysmography with SBGA (3). However, these methods transfer poorly to the study of ventilatory depression in rodent models. For example, Paco2 SBGA in rodents is a cumbersome, invasive technique requiring sophisticated surgical maneuvers. Furthermore, the pain and volume (blood) loss associated with the surgery affects the accuracy of the Paco2 measurements. An alternative means of obtaining arterial blood samples in rodents is via cardiac puncture but, like the SBGA method, lacks the ability to continuously monitor data, thereby, potentially missing transient changes. The above reasons underscore the appeal of using a noninvasive CO2 monitoring procedure for the study of induced ventilatory depression in mice. This study investigates the reliability and accuracy of assessing ventilatory depression in mice by noninvasive transcutaneous CO2 (Ptcco2) monitoring. The correlation between Ptcco2 and Paco2 measurements has been found to be high in both rats (4) and humans (5). However, previous Ptcco2 measurements in rodents have been limited to a single rat study (4). As the mouse is the rodent of choice for transgenic manipulations, we sought to expand on the rat study by validating the Ptcco2 method in the assessment of ventilatory depression in mice. Validating a simple, noninvasive method of assessing ventilatory depression in mice would go a long way towards providing a means to rapidly obtain data on numerous substances (clinically approved and experimental), under a variety of conditions and in any one of the vast number of transgenic mice that have been generated. To validate this method, we chose to induce ventilatory depression in mice by two separate means, increasing the CO2 levels in the air the mice breathed and through the administration of the opioid fentanyl. Ptcco2 and Paco2 measurements were obtained from these mice and the resulting data were correlated and subjected to Bland-Altman analysis.
Male 12- to 20-wk-old C57Bl/6 mice (Charles Rivers Hollister, CA.) were housed in groups of 8 on a 12/12-h light/dark cycle under controlled temperature with food and water provided ad libitum. All animal protocols used in this study conform to guidelines determined by the National Institutes of Health Office for Protection from Research Risks and were approved by the Animal Care and Use Committee of the Veterans Affairs Palo Alto Health Care System.
Fentanyl was obtained from Sigma-Aldrich (St. Louis, MO), dissolved in 0.9% saline as vehicle, and administered by intraperitoneal (IP) injection. Isoflurane (AErrane) was obtained from Henry Schein Inc. (Melville, NY).
We used a transcutaneous blood gases monitor and calibrator (Model 860; TCO2 M Novametrix Medical Systems Inc., Wallingford, CT.) with a CO2 Novadisc Sensor Kit (Model 6752-00) and NovaCom software. Before each experiment the probe was subjected to a two-point calibration using 5% and 10% CO2 gas standards. Calibration and measurements were performed using a constant probe temperature of 44°C. The CO2 display reading, which is by default set to reflect the metabolic factor for Ptcco2 in humans, was turned off. The readings were then corrected to 37°C by using the equation: Pco2 (37°C) = Pco2 (machine value) × exp(0.045[37°C–44°C]). The temperature correction factor of 4.5% per degree C was applied to account for the difference between probe and the blood gas analyzer (37°C) temperature.
Probe site preparation was performed on anesthetized mice 1 day before the experiment by shaving a 1 square inch area on the mouse’s abdomen with an electrical hair clipper. The following day mice were anesthetized (1.5% isoflurane in 2 L/min of O2) inside a Plexiglas box for 18–22 min to induce a stable anesthesia (anesthesia stabilization time) and transferred to a nose cone delivering 1.0% isoflurane in 2 L/min O2. Sensor contact gel was placed on the probe and fixed to the abdominal skin with adhesive ring. Throughout the experiment a stable flow of the anesthetic mixture was maintained using an Airway Gas Monitor (Model 254; Datex Instrument Corp., Helsinki, Finland) and the body temperatures of mice were maintained at 36.5°C–37.5°C using an electric heating pad. The above protocol was used in all experiments.
Arterial blood samples (0.3–0.5 mL) were collected via intracardiac puncture aimed at the left ventricle of anesthetized mice using heparinized syringes. Mice were euthanized after sample collection by placement into a CO2 inhalation chamber. Samples were immediately stored on ice and analyzed within 5 min. Paco2 was determined using i-STAT Analyzer System (Abbot, East Windsor, NJ) with temperature set at 37°C. Analyzer performance was verified daily and before each sample analysis. The disposable EG7+ cartridges were verified for calibration routinely.
As there has been no previous study of transcutaneous blood gas monitoring in mice, these experiments were performed to determine the time required to achieve stable Ptcco2 baseline values under anesthesia (anesthesia adaptation), thereby identifying the optimal time for the introduction of the CO2 gas mixture or opiate administration. A stable Ptcco2 baseline value was characterized as steady baseline Ptcco2 value (±2 mm Hg) within a 3-min interval. A group of 6 mice were tested without drug or vehicle to evaluate the average time required for anesthesia adaptation. Next, another group of 6 mice were given IP injections of saline vehicle subsequent to the anesthesia adaptation time and monitored for 100–110 min to determine the control values and the anesthetic (isoflurane) effect on the Ptcco2. Preinjection baseline Ptcco2 values acquired on each mouse were subtracted from their own postinjection values to assess the amount of Ptcco2 increase. Measurements were automatically collected by computer interface every 8 s, averaged into 32-s epochs, and graphed against time for comparison purposes. The same data management was used in all subsequent experiments.
To assure correlation between Paco2 and Ptcco2, 20 mice were randomly assigned to 3 groups A (no CO2), B (1.25% CO2), or C (2.5% CO2). All groups received 1.0% isoflurane in 2 L/min of 100% O2. After each mouse achieved a stable Ptcco2 baseline value, CO2 was supplemented for 19–21 min and blood samples collected.
(A) Naïve mice were randomly assigned into 3 groups of 6 with each mouse receiving a single IP injection of fentanyl (0.3, 0.5, or 1.0 mg/kg) after achieving a stable Ptcco2 baseline. Fentanyl doses used in this study were derived from ED50 determinations for analgesia in the tail flick assay done in-house and from the literature (6). Postinjection monitoring of Ptcco2 was performed for 70–100 min. The relationship between the increase in Ptcco2 and the dose of fentanyl was assessed.
(B) To evaluate the reproducibility of the peak increase in Ptcco2 values, 3 mice were randomly chosen from the 1.0 mg/kg fentanyl dose group for repeat measurements. Seven days after initial testing, the mice were again given the original dose of 1.0 mg/kg fentanyl and evaluated for peak increase in Ptcco2.
(C) The correlation between Paco2 and Ptcco2 measurements in drug-induced ventilatory depression was performed using 3 randomly chosen mice injected with fentanyl 1.0 mg/kg. These data were compared with the correlations previously established in Experiment 2 using CO2 inhalation. Arterial samples were obtained 20 min postinjection.
Results of Experiments 1 and 3A-B were subjected to repeated-measure one-way analysis of variance with post hoc Bonferroni’s multiple comparison test using Prism 3.0 (GraphPad, San Diego, CA). The area under the curve for experiments 3A-B was calculated by the trapezoid rule method (7). A linear regression equation was estimated from the Ptcco2 and Paco2 measurements in Experiment 2. This model was applied to the measured Ptcco2 values to obtain the derived Paco2 values. Measured and derived Paco2 values were compared using the Bland-Altman plot and the magnitude of error was determined by calculating the standard deviation of the residuals (8).
The anesthesia adaptation time required for each mouse was determined to be 35–75 min. The probe response time was found to be 30–45 s on contact in mice with the total duration of experiments being 165–190 min. The no-injection group achieved a maximum increase in Ptcco2 (mean ± sd) of 2.63 ± 2.66 mm Hg, whereas the saline-injected group displayed a maximal increase of 0.49 ± 2.74 mm Hg above baseline at the 20-min time point. There was no statistically significant difference between the two groups throughout the duration of the experiment.
All groups (A = 100% O2, B = 100% O2 with 1.25% CO2, C = 100% O2 with 2.50% CO2) reached stable Ptcco2 baseline values within 46–76 min. Group A had arterial blood drawn upon stabilization while groups B and C received the O2/CO2 mixture for 19–24 min before sampling. A good correlation between the Ptcco2 and Paco2 values was observed for mice from each group (A: Ptcco2 = [1.64 × Paco2] + 6.09; r2 = 0.89; B: Ptcco2 = [1.42 × Paco2] + 13.97; r2 = 0.93; C: Ptcco2 = [1.48 × Paco2] + 16.85; r2 = 0.86) with an overall correlation for all groups of 0.91 (Fig. 1). In view of the strength of this correlation (r2 = 0.91), a simple linear transformation was considered adequate to calculate the derived Paco2 values. To demonstrate that the differences between values obtained using the two methods remained relatively constant over the range of values a Bland-Altman analysis was performed. Bias and limits of agreement between measured and derived Paco2 values were 0.67 mm Hg and 5.68 to −4.34 mm Hg respectively (sd of bias = 2.56). The sd of the residuals (Sy.x) from the regression analysis was 2.63 mm Hg, suggesting a small magnitude of error between the transcutaneous and arterial Pco2 measurements.
(A) Fentanyl treatment: Mice used in this experiment reached baseline Ptcco2 values of 56.33 ± 6.70 mm Hg (mean ± sd) within 32–67 min of probe contact, with the total duration of these experiments lasting between 72–107 min. Different doses of fentanyl were used to assess a range of Ptcco2 responses. The 0.3 mg/kg dose produced a peak Ptcco2 response of 6.72 ± 5.05 mm Hg above baseline within 20 min; the 0.5 mg/kg and 1.0 mg/kg dose produced a peak Ptcco2 response of 15.95 ± 9.06 mm Hg and 16.18 ± 13.96 mm Hg above baseline within 20 min respectively (Fig. 2). Post hoc comparisons for each time point revealed statistically significant differences between the control and fentanyl groups (P < 0.001), as well as between the 0.3 mg/kg group and either of the 0.5 or 1.0 mg/kg fentanyl groups (P < 0.001). Further comparison of the area under the curves revealed statistically significant differences in overall group effects of treatment between the fentanyl groups and saline (F(3,30) = 20.45, P < 0.0001). Post hoc analysis found the differences between the 0.3 mg/kg fentanyl group and either of the 0.5 or 1.0 mg/kg groups to be statistically different (P < 0.01). However, the difference between the 0.5 mg/kg and 1.0 mg/kg fentanyl groups was not statistically significant.
(B) Three of the mice from the fentanyl 1.0 mg/kg dose group were retested 7–10 days later to assess within-subject reproducibility. The rapid increase to a peak Ptcco2 response again occurred within 20 min. The original mean ± sd response was 15.45 ± 0.69 mm Hg and the retested values were 16.68 ± 0.83 mm Hg above baseline. There was no significant difference between initial and retested values throughout the experiment. Further comparison of the area under the curves showed no statistically significant difference between original [976.7 (mm Hg)min] and retested [940.0 (mm Hg)min] values for the total duration of the experiment.
(C) The correlation between the Ptcco2 values and the Paco2 values obtained from mice that received fentanyl (1.0 mg/kg) were within the 95% confidence interval of Experiment 2 regression analysis curve.
A variety of factors influence Ptcco2 measurements, including probe temperature, skin thickness, presence of hair, and the anesthetic drug used (9). The effect of a heated probe on the skin is to increase metabolism of the epidermal cells (increase in Paco2) and increase capillary blood flow as a result of vasodilatation (decrease in Paco2) (10). As the living layer of skin depends partly on the outside dead layer for CO2 excretion, the skin thickness inversely affects CO2 diffusion but to a much lesser extent than O2 diffusion, because of O2 being 16 times less liposoluble (11). In rats, a probe temperature of 44°C was reported to produce the highest correlation between Paco2 and Ptcco2 (r2 = 0.924) (4). However, the authors of that study chose a probe temperature of 42°C, despite a lower correlation coefficient (r2 = 0.812), to avoid the possibility of thermal burns in juvenile animals. In the present study, using adult mice, a probe temperature of 44°C and an average probe contact time of 100 min produced only superficial indurations on the abdomen, resolving within 5 days. The ventral abdominal skin was chosen as it is thin and provides a flat surface for attachment of the probe, an important consideration given the limitations of working with a small animal. Hair removal was essential to provide a direct surface contact with the probe. Hair clipping was chosen over depilatory agents as the latter produces inflammation and alters Ptcco2 measurements (4). It has been reported that gentle application of Scotch tape removes the stratum corneum and increases the accuracy of measurements, especially in regards to transcutaneous O2 readings (12). However, we found that this method did not produce superior Ptcco2 readings in the initial experiments and was therefore discontinued.
The ventilatory depressant and hypotensive effects of the anesthetic isoflurane are dose dependant and are well documented (13,14). As suggested previously, the anesthesia stabilization and adaptation time required to reach a stable baseline was incorporated into our experimental design (4). Transcutaneous probe readings were used to assure that a stable baseline had been achieved before conducting experiments. It was determined that in the short duration of the present experiments there was only a modest increase in Ptcco2 using 1.0% isoflurane (20 min: mean increase was 0.49 ± 2.74; 60 min: mean increase was 8.50 ± 6.71 mm Hg).
Two separate approaches were used to induce ventilatory depression. The first was responsiveness to inhaled CO2, a well documented and sensitive test used in the study of ventilatory depression (15,16). The observed increase in Ptcco2 resulting from an increase in the fraction of inspired CO2 is a reliable measure of ventilatory response (17). This method is based on the physiological principal that the change in alveolar ventilation from one steady-state level to another is inversely proportional to the change in Paco2 and consequently Ptcco2. We used this principal by adding known percentages of inspired CO2 to the anesthetic mixture in Experiment 2 to assess the reliability of Ptcco2 monitoring in mice. Subsequently, we were able to validate the proportional relationship between Ptcco2 and Paco2 values in mice. The second approach used to elicit ventilatory depression was through drug induction. Paco2 and Ptcco2 values were measured in mice that had received 1 mg/kg fentanyl, an opioid that has been well documented for its ability to induce ventilatory depression. The resulting values were within the 95% confidence interval of the correlation plot generated in Experiment 2 (data not shown).
In the current set of experiments, the Ptcco2 values in mice were higher than the corresponding Paco2 values, similar to what has been observed in humans, rats, lambs, pigs, and cats (5,10). Correlating the values derived from the Ptcco2 method with values from the Paco2 method required the application of the following conversion calculation: Ptcco2 = [Constant Factor × Paco2] + adjustment factor (9). In humans, the constant factor was determined to be 1.61 with an adjustment factor of −0.01 for neonates (probe temperature 44°C) and 1.31 with an adjustment factor of 5.40 for adults (probe temperature 43°C) (5). In adult humans, the adjustment factor typically is within the range of 2–8 mm Hg, representing the local production of CO2 by skin, which is influenced by metabolism, age, and skin thickness. These factors remain constant throughout the Ptcco2 monitoring period, thereby producing consistent Ptcco2 measurements. The local skin production of CO2 is further stabilized by maintaining a constant probe temperature. Likewise, the correlation between Paco2 and Ptcco2 values in rats was reported to be Ptcco2 = [1.71 × Paco2] + 9.83 (4). The correlation equation in the present mouse study was determined to be Ptcco2 = [1.68 × Paco2] + 4.12, after temperature correction.
Fentanyl, which is used extensively in the clinical setting for the control of acute and chronic pain (18–20), was chosen as a model drug because of its well characterized ventilatory depressive effects (21,22). As fentanyl produces modest hyperthermia at the doses used in our experiments (23), we kept this variable constant by maintaining the mice at ambient body temperature of 36.5°C–37.5°C using a heating pad. Single bolus injections used in the present study mimic conditions in which fentanyl is typically administered for short-term supplementation to general anesthesia. All three doses of fentanyl (0.3, 0.5, 1.0 mg/kg) elicited a peak increase in Ptcco2 amplitude within 20 minutes postinjection. The 0.5 and 1.0 mg/kg dose elicited a similar increase in Ptcco2 peak amplitude, which was more robust than the 0.3 mg/kg dose (Fig. 2). Similarities in the peak response and slopes of the 0.5 and 1.0 mg/kg dose, support the occurrence of a ligand-specific plateau phenomenon in regards to the maximal plasma concentration of fentanyl in mice (22).
In conclusion, we have demonstrated that, under controlled conditions, Ptcco2 measurements in mice are reproducible and correlate well with Paco2 values. Moreover, a Bland-Altman analysis further demonstrated the reliability and accuracy of the Ptcco2 measurements in comparison to Paco2 measurements. These findings favorably support the use of this relatively simple, noninvasive method for reliably assessing ventilatory depression in a mammal as small as a mouse, thereby eliminating the need for specialized surgical skills. Considering the extensive number of mouse models that have been generated by gene manipulation techniques, an easily applied method for assessing ventilatory depression under a variety of conditions in any number of mouse models is of significant value. Moreover, these methods are easily applied to assessing the potential ventilatory depressive effects of a wide spectrum of clinically proven and experimental agents. Future investigations will use this Ptcco2 method for correlating the ventilatory depressive properties of a number of opioid compounds with their respective influence on in vitro activation profiles relative to the μ, δ, and κ opioid receptors (24,25). Such detailed assessments will provide the basis for rational structure-based design approaches aimed at developing the next generation of opioid receptor ligands that have adequate analgesic properties while lacking ventilatory depressive side effects (26–28).
We thank Dr. Frances Davies and Dr. Steven L. Shafer for valuable discussions throughout the course of this work.
1. Graf R, Chrubasik S, Chrubasik J, et al. Early detection of opiate-induced respiratory depression in the postoperative phase [in German]. Anaesthesiol Reanim 1995;20:38–41.
2. Bailey PL, Streisand JB, East KA, et al. Differences in magnitude and duration of opioid-induced respiratory depression and analgesia with fentanyl and sufentanil. Anesth Analg 1990;70:8–15.
3. Sandler AN, Stringer D, Panos L, et al. A randomized, double-blind comparison of lumbar epidural and intravenous fentanyl infusions for postthoracotomy pain relief: analgesic, pharmacokinetic, and respiratory effects. Anesthesiology 1992;77:626–34.
4. Stout RW, Cho DY, Gaunt SD, et al. Transcutaneous blood gas monitoring in the rat. Comp Med 2001;51:524–33.
5. Monaco F, McQuitty JC, Nickerson BG. Calibration of a heated transcutaneous carbon dioxide electrode to reflect arterial carbon dioxide. Am Rev Respir Dis 1983;127:322–4.
6. Varnado-Rhodes Y, Gunther J, Terman GW, Chavkin C. Mu opioid analgesia and analgesic tolerance in two mouse strains: C57BL/6 and 129/SvJ. Proc West Pharmacol Soc 2000;43:15–7.
7. Motulsky HJ. Prism 4 Statistics Guide-Statistical analyses for laboratory and clinical researchers. San Diego: GraphPad Software, 2003.
8. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1:307–10.
9. Severinghaus JW. A combined transcutaneous PO2
electrode with electrochemical HCO3- stabilization. J Appl Physiol 1981;51:1027–32.
10. Rithalia SV. Developments in transcutaneous blood gas monitoring: a review. J Med Engl Technol 1991;15:143–53.
11. Braems GA, Lang U, Kunzel W. Skin blood flow influences the transcutaneous PCO2
) during stress in the guinea pig. J Perinat Med 1996;24:155–61.
12. Huch R, Seiler D, Fallenstein F, et al. Use of the tcPO2
electrode for blood PO2
measurements. Birth Defects Orig Artic Ser 1979;15:573–83.
13. Steffey EP, Howland D, Jr. Comparison of circulatory and respiratory effects of isoflurane and halothane anesthesia in horses. Am J Vet Res 1980;41:821–5.
14. Klide AM. Cardiopulmonary effects of enflurane and isoflurane in the dog. Am J Vet Res 1976;37:127–31.
15. Jordan C. Assessment of the effects of drugs on respiration. Br J Anaesth 1982;54:763–82.
16. Lehmann KA, Neubauer ML, Daub D, Kalff G. [CO2
-response curves as a measure of opiate-induced respiratory depression: studies with fentanyl [in German]. Anaesthesist 1983;32:242–58.
17. Hazinski TA, Severinghaus JW, Marin MS, Tooley WH. Estimation of ventilatory response to carbon dioxide in newborn infants using skin surface blood gas electrodes. J Pediatr 1984;105:389–93.
18. Carr DB, Jacox A, Chapman CR, et al. Acute pain management: operative or medical procedures and trauma. Agency for Health Care Policy and Research. Clinical Practice Guidelines Rockville, MD: Agency for Health Care Policy and Research, 1992. AHCPR No. 95-0034.
19. Jacox AK, Carr D, Payne R, Berde CB. Management of cancer pain. Rockville, MD: Agency for Health Care Policy and Research, 1994.
20. Reisine T, Pasternak G. Opioid analgesics and antagonists. In: Goodman and Gilman’s pharmacological basis of therapeutics, 9th ed. New York: McGraw-Hill, 1997.
21. Chisholm RH, Fleischl J. Respiratory arrest with epidural fentanyl. Anaesth Intensive Care 1990;18:423.
22. Cepeda MS, Farrar JT, Baumgarten M, et al. Side effects of opioids during short-term administration: effect of age, gender, and race. Clin Pharmacol Ther 2003;74:102–12.
23. Rosow CE, Miller JM, Pelikan EW, Cochin J. Opiates and thermoregulation in mice. I. Agonists. J Pharmacol Exp Ther 1980;213:273–83.
24. Gharagozlou P, Demirci H, David Clark J, Lameh J. Activity of opioid ligands in cells expressing cloned mu opioid receptors. BMC Pharmacol 2003;3:1.
25. Gharagozlou P, Demirci H, Clark JD, Lameh J. Activation profiles of opioid ligands in HEK cells expressing delta opioid receptors. BMC Neurosci 2002;3:19.
26. Abdelhamid EE, Sultana M, Portoghese PS, Takemori AE. Selective blockage of delta opioid receptors prevents the development of morphine tolerance and dependence in mice. J Pharmacol Exp Ther 1991;258:299–303.
27. Schiller PW, Weltrowska G, Schmidt R, et al. Subtleties of structure-agonist versus antagonist relationships of opioid peptides and peptidomimetics. J Recept Signal Transduct Res 1999;19:573–88.
28. Verborgh CM, Camu F, Meert TF. Interaction between sufentanil and U-50488H with respect to antinociception and respiratory depression in rats. Acta Anaesthesiol Scand 1997;41: 895–902.