Extracorporeal Versus Conventional Cardiopulmonary Resuscitation After Ventricular Fibrillation Cardiac Arrest in Rats: A Feasibility Trial

Abstract

Objectives:

Extracorporeal cardiopulmonary resuscitation with cardiopulmonary bypass potentially provides cerebral reperfusion, cardiovascular support, and temperature control for resuscitation from cardiac arrest. We hypothesized that extracorporeal cardiopulmonary resuscitation is feasible after ventricular fibrillation cardiac arrest in rats and improves outcome versus conventional cardiopulmonary resuscitation.

Design:

Prospective randomized study.

Setting:

University laboratory.

Subjects:

Adult male Sprague-Dawley rats.

Interventions:

None.

Measurements and Main Results:

Rats (intubated, instrumented with arterial and venous catheters and cardiopulmonary bypass cannulae) were randomized to conventional cardiopulmonary resuscitation, extracorporeal cardiopulmonary resuscitation with/without therapeutic hypothermia, or sham groups. After 6 minutes of ventricular fibrillation cardiac arrest, resuscitation was performed with drugs (epinephrine, sodium bicarbonate, and heparin), ventilation, either cardiopulmonary resuscitation or extracorporeal cardiopulmonary resuscitation, and defibrillation. Temperature was maintained at 37.0°C or 33.0°C for 12 hours after restoration of spontaneous circulation. Neurologic deficit scores, overall performance category, histological damage scores (viable neuron counts in CA1 hippocampus at 14 days; % of sham), and microglia proliferation and activation (Iba-1 immunohistochemistry) were assessed.

Results:

Extracorporeal cardiopulmonary resuscitation induced hypothermia more rapidly than surface cooling (p < 0.05), although heart rate was lowest in the extracorporeal cardiopulmonary resuscitation hypothermia group (p < 0.05). Survival, neurologic deficit scores, overall performance category, and surviving neurons in CA1 did not differ between groups. Hypothermia significantly reduced neuronal damage in subiculum and thalamus and increased the microglial response in CA1 at 14 days (all p < 0.05). There was no benefit from extracorporeal cardiopulmonary resuscitation versus cardiopulmonary resuscitation on damage in any brain region and no synergistic benefit from extracorporeal cardiopulmonary resuscitation with hypothermia.

Conclusions:

In a rat model of 6-minute ventricular fibrillation cardiac arrest, cardiopulmonary resuscitation or extracorporeal cardiopulmonary resuscitation leads to survival with intact neurologic outcomes. Twelve hours of mild hypothermia attenuated neuronal death in subiculum and thalamus but not CA1 and, surprisingly, increased the microglial response. Resuscitation from ventricular fibrillation cardiac arrest and rigorous temperature control with extracorporeal cardiopulmonary resuscitation in a rat model is feasible, regionally neuroprotective, and alters neuroinflammation versus standard resuscitation. The use of experimental extracorporeal cardiopulmonary resuscitation should be explored using longer insult durations.

The invention of machines to resuscitate organisms after a period of cardiac arrest (CA) has been a topic of medical research since the 19th century (1). Cardiopulmonary bypass (CPB) was invented in 1954 (2), and has since been clinically applied, mainly in cardiothoracic surgery. Safar et al (3) pioneered the use of CPB for extracorporeal cardiopulmonary resuscitation (E-CPR) from sudden CA in animal models since 1982. In studies in dogs, the use of E-CPR enabled survival after 15–20 minutes of ventricular fibrillation (VF) CA, spontaneous circulation could be restored after up to 30 minutes (3). Many studies have used E-CPR clinically, both in adult (4–6) and pediatric patient populations (7–9). However, it is an invasive procedure and technical obstacles, such as vessel access as well as possible complications, including bleeding, air embolism, and ischemia of the extremities, have prevented its widespread use in this indication, despite encouraging results (9–11). Given the small chance to survive a CA with favorable outcome using conventional cardiopulmonary resuscitation (CPR) (12), this degree of invasiveness might be acceptable.

Large animal models (e.g., swine and dogs) of E-CPR are well established, but a rat model would add substantially to research in this field. Apart from being more economic and allowing for a larger number of experiments, rodent models introduce a wide variety of molecular and immunohistochemical tools into resuscitation research.

VF is a clinically important cardiac rhythm observed in adult CA victims. In contrast, the majority of studies of global cerebral ischemia in rodents use four-vessel occlusion techniques or CA models where the heart is stopped with asphyxia or potassium (13). Four-vessel occlusion models do not cause a systemic ischemic insult, and significant differences in resuscitation from asphyxial CA versus VF CA were identified in experimental settings (14), suggesting the need for a VF CA model.

Although survival is believed to represent an important target for the application of E-CPR, another key target is improvement in neurological outcome. Use of mild hypothermia has contributed to improved neurological outcome and reduced neuronal death in both CA and global cerebral ischemia models and in clinical VF CA (15–17). In experimental models, neuronal death, particularly delayed neuronal death in selectively vulnerable CA1 hippocampus after threshold insults, is an excellent therapeutic target to evaluate efficacy of therapies in this regard (15, 16, 18–21). However, although many drugs have been tested in experimental global cerebral ischemia and/or CA in rats, to our knowledge, the feasibility of application of E-CPR in experimental VF CA in rats and the assessment of its effect on neuronal damage, particularly in CA1 hippocampus, have not been examined. While conventional CPR improves coronary perfusion, it fails to perfuse the brain after an insult time of greater than or equal to 7 minutes (22). We hypothesized that E-CPR is feasible in rats after VF CA and improves neuropathological outcome versus conventional CPR. We used a well-described rat model of VF CA (23, 24) to assess the feasibility of E-CPR and compare it to conventional CPR. We furthermore hypothesized that E-CPR and therapeutic hypothermia can be combined in a safe and effective way.

METHODS

The study protocol was approved by the Institutional Animal Care and Use Committee of the University of Pittsburgh. Seventy adult male Sprague-Dawley rats (350–400 g; Hilltop Lab Animals, Scottdale, PA) were anesthetized with 4% isoflurane in oxygen, intubated with a 14-gauge cannula (Becton Dickinson, Sandy, UT), and mechanically ventilated (Harvard Ventilator 683, Harvard Rodent Apparatus, South Natick, MA). Anesthesia was maintained with 2% isoflurane and an FIo₂ of 0.5. Electrocardiogram (ECG), mean arterial pressure (MAP), and central venous pressure were continuously monitored and recorded (Polygraph, Grass Instruments, Quincy, MA). Rectal (Trec) was held at 37.0°C ± 0.5°C with a temperature-controlled operating table and a fan. Rats were randomized to five groups: E-CPR, E-CPR and therapeutic hypothermia (E-CPR + H), CPR, CPR + H, or sham. CPB cannulas were inserted into the right femoral artery (20-gauge Angiocath, Becton Dickinson, Sandy, UT) and the right jugular vein (five-hole 14-gauge venous cannula). In the CPR and sham groups, the right jugular vein was ligated to ensure identical surgical procedures between groups. Heparin 500 IU/kg was administered to avoid clotting of the cannulas. After surgery, FIo₂ was reduced to 0.3. VF CA was induced by a 1-minute impulse of 12 V/50 Hz alternating current and ensured by ECG readings and reduction in MAP. If spontaneous defibrillation occurred, additional 15 seconds impulses were delivered.

E-CPR Group (n = 10 for 14-Day Long-Term Outcome and Neuropathology, n = 3 for 3-Day Acute Neuropathology) and E-CPR + H Group (n = 10 for 14-Day Outcome and n = 3 for 3-Day Outcome)

The experimental timeline for the insult, initial resuscitation, and use of CPR versus E-CPR is shown in Figure 1. The use of E-CPR was based on a well-published rat model (25). The E-CPR circuit consisted of an oxygenator, an open reservoir (Ing. Martin Humbs, Ingenierbuero fuer Feinwerktechnik, Munich, Germany), tubing, and a roller pump (Masterflex, Barnant, Barrington, IL). The oxygenator contained a three-layer capillary membrane sufficient to provide a Pao₂ greater than 400 mm Hg. Temperature was controlled with a circulating water bath. The circuit was primed with a crystalloid solution (Plasma-Lyte A). Epinephrine 20 μg/kg, sodium bicarbonate 1 mmol/kg, and heparin 100 IU were added to the reservoir. At 6-minute VF CA, E-CPR was started at a flow rate of 50 mL/min, which was increased to 60 mL/min, with a mixture of 98% oxygen and 2% Co₂ at a flow of 100 mL/min to the oxygenator. Ventilation was started with an FIo₂ of 1.0 and a respiratory rate of 20 per minute. At 75 seconds of E-CPR, epinephrine 10 μg/kg was added to the reservoir. At 2 minutes of E-CPR, defibrillation was started (5 J, monophasic); defibrillation attempts were repeated every 30 seconds at 10 J for up to 5 minutes after initiation of E-CPR, then every 10 minutes for up to 1 hour. Rats were weaned from E-CPR 2 minutes after restoration of spontaneous circulation (RoSC) if the MAP was greater than 60 mm Hg.

Figure 1.

Experimental timeline, polygraph readings. Mean arterial pressures and electrocardiogram readings are presented before and during the insult and during resuscitation. CPR = cardiopulmonary resuscitation, E-CPR = extracorporeal CPR, DC = direct current defibrillation.

CPR Group (n = 10 for 14-Day Long-Term Outcome and Neuropathology, n = 3 for 3-Day Acute Neuropathology) and CPR + H Group (n = 18 for 14-Day Outcome and n = 3 for 3-Day Outcome)

Epinephrine 20 μg/kg, bicarbonate 1 mmol/kg, and heparin 100 IU were given IV 1 minute prior to start of CPR; 75 seconds after start of CPR, epinephrine 10 μg/kg was given. At 6 minutes of VF CA, ventilation (100% oxygen at prearrest ventilatory rate) and manual chest compressions (200 per min) were started. At 2 minutes of CPR, rats were defibrillated (5 J, monophasic); defibrillation attempts were repeated every 30 seconds at 10 J. If RoSC was not achieved 5 minutes after initiation of CPR, resuscitation was terminated for futility.

Sham Group (n = 10 for 14-Day Long-Term Outcome and Neuropathology)

Identical surgical procedures as in the E-CPR rats were performed in the shams, but neither CA nor CPB was used. Intensive care and anticoagulation were the same as in the CA groups.

Intensive and Intermediate Care

After RoSC, sodium bicarbonate was given to treat a metabolic acidosis, crystalloids and boluses of vasopressin 0.005 IU/kg and epinephrine 5 μg/kg were given IV to keep MAP greater than 60 mm Hg. Rats in the hypothermia group were cooled with CPB or ice for surface cooling. Rats were weaned from the ventilator at 20 minutes after RoSC and extubated 60 minutes after RoSC. Anesthesia was identical in both groups.

After decannulation, rats were placed in cages with supplemental oxygen. Activity and temperature were monitored by a Mini-mitter probe (Mini-Mitter, Sunriver, oR) inserted into the peritoneal cavity. Temperature was controlled at 37°C for 12 hours after RoSC in normothermia groups and at 33°C in hypothermia groups. The controlled temperature period was followed by spontaneous rewarming. Rats were allowed access to food and water and received subcutaneous fluid (D5W/NS 10 mL SC daily) if they did not drink. Morphine 2.25 mg was given SC if signs of distress were observed. Neurologic function was assessed daily, using a neurologic deficit score (NDS) (0% = normal, 100% = dead) and an overall performance category score (oPC) (1 = normal; 2 = slight disability; 3 = severe disability; 4 = comatose; 5 = dead).

On day 3 or 14, after assessment of final oPC and NDS, rats were killed with an isoflurane overdose and perfused with normal saline followed by 10% formalin for histologic evaluation. Additional rats in the E-CPR, E-CPR + H, CPR, and CPR + H groups were assessed (n = 3 in all groups), which survived for 3 days, to compliment the definitive long-term neuropathological studies and provide insight into the time course of neuronal death and microglia proliferation and activation in this model.

Histopathology

Formalin-fixed brains were divided into eight coronal slices 3 mm apart and embedded in paraffin blocks. Five micromolar sections were stained with hematoxylin and eosin (H&E) and with Fluoro-Jade C (FJC) (Millipore, Temecula, CA)—the latter stain according to the procedure of Schmued et al (26). Sections were also stained with a rabbit antibody to Iba-1 (Wako Chemicals, Richmond, VA) to immunohistochemically identify, quantify, and characterize microglia. Sections were rehydrated and blocked for endogenous peroxidases. After antigen retrieval with citrate buffer, sections were blocked for 30 minutes in 3% normal goat serum. Sections were incubated overnight at 4°C with a 1:250 dilution of the primary anti Iba-1 antibody, then with a goat anti-rabbit secondary antibody (1:200), washed, and incubated with ABC peroxidase kit (Vector, CA). 3,3′-diaminobenzidine (Vector) was used for visualization. Sections were counterstained with hematoxylin. Dr. Janata was presented histologic photographs and counted viable neurons in CA1 blinded to the group assignment. With the median of viable neurons in the sham group defined as 100%, the number of viable neurons in the CA1 region was calculated and is presented as the percentage of sham. The microglial response was also quantified in CA1 hippocampus by an observer blinded to group assignment. Specifically, both microglial activation and proliferation were evaluated. As previously described, a semiquantitative score (0–3) was used to evaluate microglial activation (27). Activated microglia with ameboid cell bodies and thick ramified processes were then counted in the midsection of the CA1 region by Dr. Drabek masked to the treatment (28). Microglia scores are expressed as median (range) and microglial counts as mean ± sd.

In addition, neuropathological changes, including neuronal degeneration, microglial activation, and reactive astrocytosis in multiple brain regions, were evaluated qualitatively by Dr. Garman. A neuropathological rating scale from 0 (no damage) to 5 (maximal damage) was similarly quantified by Dr. Garman for both H&E- and FJC-stained sections. The various brain regions assessed in this study are depicted in Figure 2.

Figure 2.

Brain regions examined for both quantitative assessments (CA1 hippocampus) and qualitative scoring (all other regions). Please see text for details of methods and results in the various brain regions. med = medial; dors = dorsal; post/vent = posterior ventral.

Statistical Analysis

Continuous data that were normally distributed were reported as mean and standard deviation, data that were not normally distributed were reported as median and range or median and 25th/75th percentile. Group comparisons were made with one-way analysis of variance and the Student-Neuman-Keuls test for pair-wise analysis, or the Kruskal-Wallis test, as appropriate. Categorical variables were reported as counts, group comparisons were made with the chi-square test. our significance level was a two-sided p value of less than 0.05. All calculations were performed with PASW Statistics for Mac 18.0 (SPSS, Chicago, IL).

RESULTS

Rats weighed 394 ± 16 g in the E-CPR group, 382 ± 10 g in the CPR group, 398 ± 21 g in the E-CPR + H group, 379 ± 5 g in the CPR + H group, and 386 ± 14 g in the sham group (p = not significant). Baseline physiologic variables are presented in Table 1.

Table 1.

Physiologic Variables at Baseline and 1 hr After Restoration of Spontaneous Circulation

	Extracorporeal Cardiopulmonary Resuscitation n = 10	Cardiopulmonary Resuscitation n = 10	Extracorporeal Cardiopulmonary Resuscitation and Therapeutic Hypothermia n = 10	Cardiopulmonary Resuscitation and Therapeutic Hypothermia n = 18	Sham n = 10
Physiologic variables at baseline
Heart rate (min−1)	355 ± 26	356±14	361 ± 12	353 ± 15	359 ± 26
Mean arterial pressure (mm Hg)	102 ± 9	101 ±11	91 ± 4	94 ± 4	98 ± 7
Rectal	36.8 ± 0.3	36.9 ± 0.3	37.3 ± 0.4	37.1 ± 0.3	37.1 ± 0.4
PH	7.38 ± 0.03	7.39 ± 0.02	7.38 ± 0.02	7.41 ± 0.04	7.36 ± 0.03
Pao2 (mm Hg)	136 ± 37	126 ± 25	115± 13a	163 ± 39	123 ± 33
Paco2 (mm Hg)	37 ± 4b	38 ± 3	36 ± 5	37 ± 3	42 ± 7
Hematocrit (%)	35 ± 2	36 ± 3	37 ± 4	37 ± 3	35 ± 2
Base excess (mEq/L)	−2 ± 2	−1 ± 1	−3 ± 2	−1 ± 2	−1 ± 3
K (mmol/L)	3.7 ± 0.3c	4.2 ± 0.3	3.7 ± 0.6	3.9 ± 0.4	4.2 ± 0.4
Na (mmol/L)	141 ± 2	140 ± 2	147 ± 8	143 ± 3	139 ± 2
Glucose (mg/dL)	219 ± 34	219 ± 45	242 ± 18	258 ± 60	231 ± 74
Lactate (mmol/L)	1.6 ± 0.4	1.1 ± 0.8	1.8 ± 0.5	1.2 ± 0.4	1.7 ± 0.9
	Extracorporeal Cardiopulmonary Resuscitation n = 10	Cardiopulmonary Resuscitation n = 8	Extracorporeal Cardiopulmonary Resuscitation and Therapeutic Hypothermia n = 9	Cardiopulmonary Resuscitation and Therapeutic Hypothermia n = 13	Sham n = 10
Physiologic variables at 1 hr after restoration of spontaneous circulation
Heart rate (min −1)	371 ± 27	338 ± 23d	327 ± 24	332 ± 48	373 ± 23
Mean arterial pressure (mm Hg)	93 ± 12	99 ± 19	79 ± 16a	99 ± 10	107± 14
Rectal	371 ± 0.5	36.3 ± 0.6	33 ± 0.3a	34 ± 0.6	37.0 ± 0.5
pH	7.46 ± 0.04	7.46 ± 0.03	7.43 ± 0.10	7.38 ± 0.06	7.41 ± 0.05e
Pao2 (mm Hg)	218 ± 95f	117 ± 45f	414 ± 38a	186 ± 61	341 ± 72f
Paco2 (mm Hg)	32 ± 4	32 ± 4	38 ± 10	42 ± 4	36 ± 6
Hematocrit (%)	32 ± 2	33 ± 3	35 ± 1	41 ± 5	34 ± 2
Base excess (mEq/L)	0 ± 2	1 ±1	0 ± 1	0 ± 3	1 ± 4
K (mmol/L)	4.0 ± 0.6	3.7 ± 0.4g	3.3 ± 0.1	2.8 ± 0.5	4.3 ± 0.4
Na (mmol/L)	145 ± 3	148 ± 3	153 ± 3	142 ± 23	141 ± 2e
Glucose (mg/dL)	165 ± 46	175 ± 74	282 ± 28	196 ± 71	161 ± 51
Lactate (mmol/L)	5.4 ± 2.5c	3.2 ± 1.0	8.0 ± 2.8	3.7 ± 3.0	1.9 ± 1

Hypothermia groups:

^ap < 0.05 extracorporeal cardiopulmonary resuscitation and therapeutic hypothermia vs cardiopulmonary resuscitation and therapeutic hypothermia, Normothermia groups,

^bp < 0,05 comparing extracorporeal cardiopulmonary resuscitation vs sham,

^cp < 0,05 comparing extracorporeal cardiopulmonary resuscitation vs cardiopulmonary resuscitation and sham,

^dp < 0,05 comparing cardiopulmonary resuscitation vs extracorporeal cardiopulmonary resuscitation and sham,

^ep < 0,05 comparing sham vs extracorporeal cardiopulmonary resuscitation and cardiopulmonary resuscitation,

^fp < 0,05 for all groups,

^gp < 0,05 comparing cardiopulmonary resuscitation vs sham,

Data are given as mean ± sd, Data are presented at baseline and 1 hr after restoration of spontaneous circulation,

Resuscitation

Outcome data are given in Table 2. Seven rats did not achieve RoSC; the animals that did not achieve RoSC were 0 of 10 in the E-CPR group, 1 of 10 in the CPR group, 1 of 10 in the E-CPR + H group, and 5 of 18 in the CPR + H group. There were no significant group differences in RoSC rates. There were no differences in requirements of epinephrine or sodium bicarbonate after RoSC. Physiologic data of 1 hour after RoSC are presented in Table 1. The full time course of the key variables MAP, heart rate, and Trec during resuscitation and intensive care is shown graphically in Figure 3, A–C. E-CPR led to a more rapid induction of hypothermia than in the other groups (p < 0.05), although heart rate was lowest in the E-CPR + H group (p < 0.05). MAP at 5 minutes was significantly lower in the CPR group versus sham. At 15 minutes, MAP was lowest in E-CPR + H (p < 0.05), for the remaining observation period, it was not different between E-CPR + H and E-CPR.

Table 2.

Outcome in Terms of Final Overall Performance Categories (1–5) at 14 Days

	Extracorporeal Cardiopulmonary Resuscitation	Cardiopulmonary Resuscitation	Extracorporeal Cardiopulmonary Resuscitation and Therapeutic Hypothermia	Cardiopulmonary Resuscitation and Therapeutic Hypothermia	Sham
Overall performance category 1	•••	•••	•••	•••	•••••
	•••	••	••	••	•••••
overall performance category 2	•
overall performance category 3			•	•
Overall performance category 4
Overall performance category 5	•••	••••	•••	••••
				•••
Neurologic deficit scores (mean ± standard deviation)	2 ± 3	1 ±1	1 ± 2	1 ± 2	0 ± 0

Each dot represents one rat.

CA1 (viable neurons in the CA1 region of the hippocampus, % of sham, median, and interquartile range).

Figure 3.

Rectal temperature (A), mean arterial pressure (B), and heart rate (C) from baseline to 60 min after restoration of spontaneous circulation. Extracorporeal cardiopulmonary resuscitation (E-CPR) led to a more rapid induction of hypothermia than in the other groups (p < 0.05), although heart rate was lowest in the E-CPR hypothermia group (p < 0.05). p < 0.05 is indicated by brackets and further explained by italic letters: a, cardiopulmonary resuscitation (CPR) versus sham; b, extracorporeal cardiopulmonary resuscitation plus hypothermia (E-CPR + H) vs all other groups; c, E-CPR + H vs sham; d, E-CPR + H vs E-CPR and sham; e, E-CPR + H vs cardiopulmonary resuscitation plus hypothermia (CPR + H), CPR, and sham; f, E-CPR, CPR + H, and CPR vs sham; g, CPR + H vs E-CPR and sham; h, E-CPR + H and CPR + H vs E-CPR and sham; i, CPR + H vs E-CPR, CPR, and sham; j, E-CPR and CPR vs sham. Please see text for details. BL = baseline.

Outcome (Survival, OPC, and NDS)

In the sham group, 10 of 10 rats survived without neurological damage. Favorable gross neurological outcome (oPC 1 or 2) was observed in 7 of 10 rats in the E-CPR group, 5 of 9 rats in the CPR group, 5 of 9 rats in the E-CPR + H group, and 5 of 13 rats in the CPR + H group. NDS and oPC are shown in Table 2. There was no difference between E-CPR and CPR groups with or without hypothermia on any of the days.

CA1 Hippocampal Neuronal Death

Representative images from coronal brain sections taken through the dorsal hippocampus are shown in Figure 4. There was a highly significant decrease in the number of surviving CA1 hippocampal neurons as assessed by H&E between 3 days and 14 days after CA as depicted in Figure 5 (p = 0.000 for 3 d vs 14 d by Mann-Whitney test). The median value of the number of surviving neurons ranged from 22% to 39% of sham neuron counts in CA1 at 14 days after CA, but these values did not differ between CPR versus E-CPR and hypothermia versus normothermia treatments.

Figure 4.

Representative coronal hippocampal CA1 sections from 14- and 3-d outcome animals. Hematoxylin and eosin staining, magnification ×10; Fluoro-Jade C staining for neurons undergoing cell death (bright green), magnification 4×; anti-Iba-1 staining for activated microglia (brown), magnification ×20. CPR = cardiopulmonary resuscitation; E-CPR = extracorporeal cardiopulmonary resuscitation.

Figure 5.

Surviving CA1 hippocampal neurons quantified in hematoxylin and eosin-stained coronal brain sections in the various treatment groups at 3 and 14 d after the insult. overall neuronal survival was significantly reduced between 3 and 14 d (*p < 0.05); however, there was no significant effect of treatment within either the 3 d or 14 d groups. CPR normo = cardiopulmonary resuscitation plus normothermia; E-CPR normo = extracorporeal cardiopulmonary resuscitation plus normothermia; CPR-hypo = cardiopulmonary resuscitation plus hypothermia; E-CPR-hypo = extracorporeal cardiopulmonary resuscitation plus hypothermia.

Neuropathologic Damage Scores

As outlined, neuropathological damage scores were generated for both H&E-stained sections and FJC-stained sections in nine separate brain regions for each rat at 14 days after injury. There were no significant differences between normothermic CPR and E-CPR groups in any brain region at 14 days after CA (Table 3). In contrast, hypothermia significantly reduced neuropathological damage scores in the subiculum (p < 0.05 for E-CPR hypothermia vs both CPR-normothermia and E-CPR normothermia, and for CPR-hypothermia vs E-CPR normothermia), Kruskal-Wallis test and Dunn post hoc comparison corrected for multiple comparisons) in FJC-stained sections at 14 days after the insult (Fig. 6). In addition, hypothermia attenuated neuronal death in ventral thalamus at 14 days after CA (p = 0.015 for all hypothermia treated vs all normothermia treated rats as assessed by H&E; Fig. 7) and across the entire thalamus (ventral and dorsal) when comparing all hypothermia treated versus all normothermia treated rats at 14 days (p = 0.039 again as assessed by H&E). There were no differences between hypothermic and normothermic groups for neuropathological damage scores by either H&E or FJC in any of the other six brain regions at 14 days after the insult (Table 3).

Table 3.

Neuropathologic Damage Scores at 14 Days After Cardiac Arrest

Brain region	Cardiopulmonary Resuscitation Normothermia		Extracorporeal Cardiopulmonary Resuscitation Normothermia		Cardiopulmonary Resuscitation Hypothermia		Extracorporeal Cardiopulmonary Resuscitation Normothermia
	Hematoxylin and Eosin	Fluoro-Jade C	Hematoxylin and Eosin	Fluoro-Jade C	Hematoxylin and Eosin	Fluoro-Jade C	Hematoxylin and Eosin	Fluoro-Jade C
CA1	4 (2.75–5)	4 (2.75–5)	4.5 (3.25–5)	4.5 (3.25–5)	3 (3–3.75)	3 (3–3.75)	4 (3–5)	4 (3.25–4)
Striatum	0 (0–0.25)	0 (0–0)	0.5 (0–1)	0.5 (0–1)	0 (0–0)	0 (0–0)	0.5 (0–1)	1 (0.25–1)
Cortex	0 (0–0.50)	0 (0–0.50)	0 (0–0)	0 (0–0)	0 (0–0)	0 (0–0)	0 (0–0.75)	0 (0–0)
Medial geniculate	0 (0–1)	0 (0–0)	0 (0–0.75)	0 (0–0)	0 (0–0.75)	0 (0–0)	0 (0–0)	0 (0–0)
Substantia nigra	1 (0–3)	0 (0–0)	0 (0–1.5)	0 (0–0)	0 (0–1.5)	0 (0–0)	0.5 (0–1.75)	0 (0–0)
Purkinje cells	0 (0–0.50)	0 (0–0.50)	0 (0–0)	0 (0–0)	0 (0–0)	0 (0–0)	0 (0–0.75)	0 (0–0)

Values are median (25th to 75th percentile).

Figure 6.

Neuropathological damage scores as assessed using Fluoro-jade C (FJ-C) staining in the subiculum across treatment groups at 14 d after the insult. Hypothermia significantly reduced neuropathological damage scores in the subiculum (*p < 0.05 for extracorporeal cardiopulmonary resuscitation plus hypothermia [E-CPR hypo] vs both cardiopulmonary resuscitation plus normothermia [CPR normo] and extracorporeal cardiopulmonary resuscitation plus normothermia [E-CPR normo], and for cardiopulmonary resuscitation plus hypothermia [CPR-hypo] vs E-CPR normothermnia). Please see text for details.

Figure 7.

Neuropathologic damage scores as assessed using hematoxylin and eosin staining in the thalamus across treatment groups at 14 d after the insult. Hypothermia significantly attenuated damage in ventral thalamus at 14 d after cardiac arrest (*p = 0.015 for all hypothermia treated vs all normothermia treated rats as assessed by hematoxylin and eosin [H&E]) and across the entire thalamus (ventral and dorsal) when comparing all hypothermia treated vs all normothermia treated rats at 14 d (p = 0.039). Please see text for details. CPR normo = cardiopulmonary resuscitation plus normothermia; E-CPR normo = extracorporeal cardiopulmonary resuscitation plus normothermia; CPR-hypo = cardiopulmonary resuscitation plus hypothermia; E-CPR-hypo = extracorporeal cardiopulmonary resuscitation plus hypothermia.

Iba-1 Immunohistochemistry

Figure 4 also shows representative Iba-1 immunostaining in CA1 to identify the microglial response to CA at 3 and 14 days after the insult. Iba-1 staining was prominent in neuroanatomic regions where cellular degeneration was evident with H&E and FJC, being most pronounced in CA1, in the thalamic reticular nuclei, and the ventral posterolateral thalamic nuclei. In shams, few individual scattered quiescent microglia with small soma and long processes were observed. There were no activated microglia present. There was only mild microglial activation and proliferation observed at 3 days, similar in all CA groups irrespective of temperature management or the resuscitation method. Both microglia activation and proliferation increased over time from 3 days to 14 days. At 14 days, there was a highly significant increase in microglial activation (score) and proliferation (count) versus sham (Fig. 8, A and B). The resuscitation method (CPR vs E-CPR) had no effect on the microglial response. Surprisingly, overall microglial activation and proliferation were more pronounced in CA1 at 14 days in the hypothermic versus normothermic groups (CPR and E-CPR groups combined, p < 0.05 for both microglial score and count; Fig. 8, A and B).

Figure 8.

Microglial response to cardiac arrest (CA) in CA1 hippocampus as assessed by immunohistochemical detection of Iba-1. CA produced a dramatic microglial response in the hippocampus (*p < 0.001 shams vs other groups). Therapeutic hypothermia (Hypo) resulted in increased microglial activation (A, microglial score) and proliferation (b, microglial count) at 14 days (both **p < 0.05 for normothermia [Normo] vs hypothermia treated rats). The resuscitation method (i.e., CPR vs E-CPR) did not affect the microglial response, and thus those groups were combined to specifically examine the effect of hypothermia.

Neuropathology

Microscopic findings for both 3-day and 14-day surviving rats were similar except that the intensity of FJC staining was often diminished in rats after 14 days versus the same regions in 3-day rats, both in the normothermia and hypothermia groups and for the progressive neuronal loss in CA1 by day 14 seen on H&E—as previously discussed. Indeed, the greatest degree of degeneration was present in CA1 in all rats (Fig. 4). In the thalamus, particularly the ventral posterolateral and reticular nuclei and, to a lesser extent, the dorsal nuclei were affected; other thalamic regions sometimes found to have foci of degeneration included the medial geniculate bodies, the ventroposteromedial nucleus, and the lateral geniculate body. Neuroanatomic regions showing less consistent lesions included the cerebral cortex, the striatum, and the globus pallidus. The substantia nigra pars reticularis was injured in two rats in the CPR group and four rats in the E-CPR group. The subiculum appeared to be altered in H&E-stained sections and was consistently stained with FJC. However, only the neuropil stained with FJC, which may represent degenerative processes extending into the subiculum from neurons in CA1.

DISCUSSION

To our knowledge, this is the first study using E-CPR in a rat model of VF CA. It showed that E-CPR is effective and feasible for this indication. Neurologic outcome was favorable in all surviving rats. However, histological damage was not attenuated by E-CPR and damage was most pronounced in CA1 and not different between E-CPR and conventional CPR groups. Adding mild hypothermia to both resuscitation techniques did not significantly attenuate CA1 damage. Damage was also widespread and detectable in other brain regions, notably the thalamus after either E-CPR or conventional CPR resuscitation (Figs. 6 and 7 and Table 3). Hypothermia, however, reduced damage in the subiculum and the thalamus. Nevertheless, there was no synergistic benefit when E-CPR was combined with hypothermia in our paradigm. Hypothermia also had a surprising effect increasing the microglial response in CA1 at 14 days after the insult.

We suggest that both the E-CPR and the CPR groups in our 6-minute model appear to be attractive control groups for future studies of neuroprotective therapies, with histological damage and/or mortality as key outcome variables. Reperfusion with E-CPR improved neither neurologic nor histologic outcome versus CPR. The failure to attenuate CA1 neuronal death or other neuropathological damage using E-CPR versus conventional CPR suggests that initial reperfusion with CPB versus CPR. The failure to attenuate CA1 neuronal death or other neuropathological damage using E-CPR versus conventional CPR suggests that initial reperfusion with CPB versus chest compressions and more rapid cooling on bypass versus surface cooling was not sufficient to prevent neuronal death. This could be explained by several factors, such as: 1) the short duration of CA in healthy rats, which required a relatively brief resuscitation efforts in either group, 2) the relatively short duration of mild hypothermia (12 hr) that we used after RoSC, 3) the need for highly specific neuron-targeted therapies to block intracellular death cascades that are set into motion after even brief insults, such as a 6-minute arrest, and/or 4) the lack of secondary insults that might occur clinically in patients with severely compromised myocardial function not supported with E-CPR—as might be seen with clinical application of E-CPR. Arterial po₂ was lower in the CPR group, potentially indicating trauma to the lungs during chest compressions; however, the impact of various levels of Pao₂ after CA is controversial (29). Two cases of infection support the need for antibiotic prophylaxis in future studies on E-CPR in rats.

We observed mortality in both groups. It is well recognized that mortality after VF CA in humans is substantial. Furthermore, the effect of the resuscitative drugs epinephrine and bicarbonate, used in our study, remains controversial with both beneficial (30) and detrimental (31, 32) effects reported after CA in rats.

Our aims were both to determine if E-CPR conferred benefit over standard resuscitation after a threshold VF insult and to expand our understanding of the neuropathology after VF CA in rats and create a map of the temporal and regional extent of brain damage in this model. At 2 weeks of recovery, damage in the CA1 region was most pronounced, but additional foci of damage were found. Histopathologic changes in thalamic nuclei and subiculum, among other brain regions, suggest that these neuroanatomic regions should be evaluated in future studies of global ischemia and CA. It is possible that neuronal death in various brain regions will require specific therapies so regional assessment of damage could be important. We were surprised that significant injury to the striatum and Purkinje neurons in the cerebellum were not observed in light of the density of damage in CA1 (33). Nevertheless, given that damage is not limited to CA1 after only a 6-minute insult, injury outside the hippocampus may take on even greater significance after more prolonged insults. Finally, as assessed by H&E, we did not see statistically significant neuronal death in CA1 at 3 days after the 6-minute VF insult in any group. In contrast, neuronal cell loss in CA1 was robust on H&E at 14 days. Thus, we did not carry out separate sham rats at 3 days. Also given that the sample size in the 3-day outcome groups was small, we focused on neuropathology in those rats rather than behavioral outcomes that are more definitively characterized in the 14-day outcome studies.

Precise and rapid temperature control was achieved with the use of CPB. Despite this, we found no significant effect of hypothermia on mortality, no benefit on CA1, while protection was conferred in the subiculum and the posterior and ventral thalamus. Full protection of CA1 was achieved in the six animals treated with hypothermia that were killed 72 hours after the insult, and this is similarly reflected in the images in Figure 4. However, we did not study a large sample size at this early postresuscitation time point given the well-known evolution of neuronal death that occurs after global cerebral ischemic insults to the brain (34). In this model, it seemed that hypothermia did not prevent, but delayed damage in CA1. This has implications for VF CA outcome studies in rodents. Choosing a short observation period might lead to a misinterpretation of data. We used therapeutic hypothermia for a duration of 12 hours. A longer treatment, such as 48 hours, might have yielded more benefit based on the seminal work of Colbourne et al (15, 35).

Comparison of H&E- and FJC-stained sections from the 3- and 14-day rats was informative regarding the determination of optimal time points for demonstrating neuronal degeneration with either stain. At the 3-day time point, degenerating cells sometimes appeared viable on H&E, whereas FJC already marked them as moribund. on the other hand, foci of degeneration identified in H&E-stained sections by the presence of hypercellularity and/or neuropil vacuolization were frequently devoid of FJC staining in the 14-day outcome rats. The reason for discordance between the findings in the H&E- and FJC-stained sections is likely due to the fact that large-sized neurons (such as the pyramidal neurons within Ammon’s horn of the hippocampus) tend to become mummified after degeneration and, therefore, are visible for longer periods of time than do smaller-sized neurons that frequently disappear after only a few days after degeneration. It is a limitation of the study that there were only three rats in each of the 3-day outcome groups. However, we chose to maximize on the more relevant 14-day time point. our histological findings mirror those previously described after asphyxial CA in rats (36). Similar results were observed by others after VF CA in rats (37).

Microglia activation as identified by Iba-1 staining was also consistently observed, being most marked in CA1, colocalized with the region of extensive neuronal death. Microglia could be an independent determinant of neurologic outcome. However, we could not determine if microglia proliferation appeared secondary to neuronal death and was associated with removal of cell debris, neuronal recovery, or plasticity, or if these cells mediated secondary injury. Nevertheless, our findings suggest that microglial cells are being recruited over time to the regions of neuronal degeneration. The use of E-CPR did not affect the microglial response. Surprisingly, therapeutic hypothermia resulted in a more pronounced microglial activation and proliferation at 14 days. This delayed, rather than early, appearance after brain ischemia has been previously described (38) and argues against a major role of microglia as a death effector early after the insult in our model. Similarly, Faustino et al (39) recently reported that depletion of microglia in a model of focal brain ischemia in neonatal rats increased the amount of damage at 72 hours after reperfusion, suggesting a subacute beneficial effect of microglia—and consistent with our findings—i.e., some degree of neuroprotection by hypothermia, and a magnified microglial response. Given the robust microglial response, our model appears suitable for testing drugs modifying microglia, to shed light on whether their role after CA is detrimental, reparative, or an epiphenomenon.

While neuropathologic changes were consistent, NDS and oPC scores were unable to detect neurological sequelae 14 days after the experiment. Use of more sensitive behavioral tests tailored for VF CA is therefore essential. Tests of cognitive function, such as the Morris water maze or fear conditioning, widely used to quantify changes after traumatic brain injury, should be adapted to our VF CA model; those studies are ongoing.

E-CPR in a rat CA model was first studied by Han et al (40) in an asphyxial CA model. Rats were subjected to 8 minutes of asphyxial CA and resuscitated with E-CPR. Spontaneous RoSC occurred after 47 ± 25 seconds of reperfusion. our VF model adds to asphyxial CA models the possibility to study E-CPR without a beating heart, as RoSC did not occur spontaneously. E-CPR without concurrent spontaneous heartbeat is likely to be a major application of E-CPR in CA patients. As E-CPR induces a nonpulsatile flow, critical organ perfusion might be different without a spontaneous heartbeat. Our work is highly germane to CA in adults, most of which results from VF or other arrhythmias rather than asphyxia.

The applicability of our results to a possible use of E-CPR in human CA victims is limited due to our experimental design for this initial study, which aimed at achieving survival in both CA groups so that neuropathology could be compared. We used a CA duration of only 6 minutes, while the preferred indication for E-CPR in clinical trials is prolonged or refractory CA, where few survivors in control groups using chest compressions would be expected (41, 42). Our study used E-CPR without taking maximal advantage of some of its potential benefits. Extracorporeal cooling is one of the most effective ways to induce hypothermia after CA (10, 43, 44), and cooling was faster in our E-CPR hypothermia versus CPR hypothermia group. However, this might not have conferred a maximal advantage in rats (unlike humans) that can be rapidly cooled with surface methods. Finally, VF CA does not result from myocardial infarction in our otherwise healthy rats, and hemodynamic support by E-CPR in that setting could contribute importantly to its benefit, a situation that was not modeled in our study.

CONCLUSIONS

E-CPR after a 6-minute VF CA in a rat model is feasible, but did not improve neurologic or histological outcome versus CPR, possibly due to short durations of the insult and resuscitation efforts. Most survivors achieved full recovery of gross neurological function on OPC and NDS. However, extensive neuronal death was present after either E-CPR or conventional CPR in multiple brain regions, associated with microglial proliferation. Although a brief 12 hours application of mild hypothermia was used in this model, beneficial effects were region dependent, seen in subiculum and thalamus, but there was no synergistic benefit from the combination of E-CPR and mild hypothermia. Hypothermia also produced an augmentation of the microglial response, suggesting a possible beneficial role of microglia after CA. Further studies of E-CPR after VF CA are warranted using our new paradigm, using longer insult times and longer durations of therapeutic hypothermia.