Coronary versus carotid blood flow and coronary perfusion pressure in a pig model of prolonged cardiac arrest treated by different modes of venoarterial ECMO and intraaortic balloon counterpulsation

Under fluoroscopic guidance, a coronary 6 F AR1 guiding catheter (Cordis, Miami, Fl, USA) was placed into the ostium of the left main coronary artery and a 6 F Headhunter carotid access or RCB guiding catheter (Cordis) was placed directly into the left carotid artery, or through the bicarotid trunk whenever present, 1 to 2 cm behind the ostium. An IABP balloon was inserted into the descending aorta and Swan-Ganz and coronary sinus catheters were advanced under fluoroscopic control.

The right subclavian artery was surgically exposed and a 15 F ECMO cannula was inserted with its tip positioned to the origin of the right subclavian artery for later extracorporeal circuit connection. The cannula was clamped and continuously flushed with heparinized saline to avoid thrombosis. Thereafter, Doppler flow wires were inserted through guiding catheters to be positioned approximatelly 3 to 5 cm above the origin of the left carotid artery and in the proximal straight segment of the circumflex coronary artery, see Additional file 2. The Doppler wires were carefully manipulated and placed for obtaining the best possible ultrasound signals. The protocol consisted of 15-minute intervention intervals outlined in Figure 1. Following stabilization, all hemodynamic and flow values in each of these intervals were obtained every five minutes and averaged, the blood was always drawn at the end of each interval. The baseline interval was followed by the IABP set to a 1:1 mode to adjust for the best possible augmentation. Thereafter, the ECMO cannulae were inserted according to randomization: in the FF arm, both venous and arterial femoral cannulae; in the case of the FS arm, only a venous femoral cannula was inserted and the previously prepared subclavian cannula was used. Fluoroscopically, femoral artery ECMO cannula position was controlled not to be in close proximity to the IABP balloon to avoid interference with aortic ECMO flow. Initially, a basal mode of < 10 mL/kg/minute blood flow was set for stabilization. Next, a maximal ECMO flow mode was defined by a blood flow rate of 100 to 130 mL/kg/minute. Then, cardiac arrest was commenced by VF induction using programmed ventricular stimulation and ECMO flow was set to 5 to 10 mL/kg/minute. After 15 minutes of cardiac arrest, the ECMO flow was increased to reach the target flow of 100 mL/kg/minute as soon as possible. After finishing the ECMO interval and all data retrieval, IABP was switched on to an internal mode of 100 inflations/minute. Then, an ECMO switch procedure was carried out and all values were again retrieved from ECMO and ECMO + IABP configurations. Vasopressors were not used in these ECMO-treated phases of cardiac arrest. After all values and blood specimens were obtained, standard CPR including chest compressions, whenever needed, were started with 270 J biphasic DC defibrillation (TEC-550, Nihon Kohden, Japan). Return of spontaneous circulation (ROSC) was evaluated based on the following definition: a supraventricular rhythm with hemodynamically effective pulsations regardless of ECMO flow with mean arterial pressure of 60 mmHg. The evaluation took place at 5 and 60 minutes following first defibrillation. An ECMO weaning trial was performed regularly and until 60 minutes post CPR, hemodynamic tolerance was tested. Hypotension was eventually corrected with standard norepinephrine dosing administered continuously by an IV drip. Thereafter, the animal was euthanized by morphine and propofol overdose followed by intravenous potasium chloride 1 mmol/kg.

Descriptive data are presented as median (interquartile range) of measured values or as median (interquartile range) of individual baseline percentages of each experiment. The median of baseline percentage is not percent expression of absolute value median. It is a median of baseline percentage from each experiment. Repeated analysis of variance measures were used to assess the between-subject effect within each arm of the study (that is, we assessed the effect of the ordering of ECMO procedures as one arm started with FF ECMO, whereas the other started with FS ECMO; Figure 1). No difference was noticed. We then pooled the corresponding methods from both arms to assess the differences. Each animal was used as its own control, and paired comparisons were made between respective approaches. A Friedman test and Tukey's post hoc analysis were performed for different analyses of ECMO approaches. P-values < 0.05 were considered statistically significant. All statistical analyses were performed by MedCalc software version 11.6. (MedCalc Software, Broekstraat 52, 9030 Mariakerke, Belgium) and MATLAB Statistics Toolbox (MathWorks Inc., Natick, MA, USA).

Carotid and coronary blood flow velocities presented as average peak velocities are enumerated in Table 1. Absolute values give velocities at baseline, cardiac arrest and different ECMO regimens. Cardiac arrest values of 12 (7.0, 12.0) cm/sec for carotid and 8 (7.0, 9.8) cm/sec for coronary flow velocities reflect very low flows generated by basal ECMO setting to prevent thrombosis within the ECMO circuit and simulating a "low-flow" cardiac arrest state. Relative carotid and coronary blood flow velocities according to different ECMO regimens during the continued cardiac arrest treated by ECMO are graphically outlined with respective statistical evaluation in Figure 2.

Carotid flow velocities in all ECMO approaches regardless of IABP presence reached values higher than 80% of baseline flow and were near to 100% of baseline values in both FS and FS + IABP configurations. Addition of IABP to FF ECMO did not change the carotid flow significantly.

Coronary flow velocities are lower, reaching about a half of the carotid velocities. Only the FF regimen assures values over 80% of baseline. Addition of IABP to the FF approach caused a statistically significant decrease in coronary flow velocity, P = 0.004. The FS ECMO approach reached only 70.0% (49.1, 113.2) of the baseline flow, which is significantly lower than 90.0% (66.1, 98.6) of FF ECMO, P = 0.039. Addition of IABP to FS did not significantly change the coronary blood flow velocity. Comparison of both IABP regimens shows a statistically significant difference, FS +IABP vs. FF + IABP 76.7% (71.9, 111.2) vs. 60.7% (55.1, 86.2), P = 0.026. Thus, the FF+IABP approach reached the lowest coronary flow velocities in this cardiac arrest setting.

Brain and peripheral regional oxygen saturations are demonstrated in Table 2, and the values relative to baseline during different ECMO regimens and statistical evaluation are outlined in Additional file 3. Both brain and peripheral regional oxygen saturations dramatically decreased immediately after induction of VF cardiac arrest to 23.0% (15.0, 32.3), P = 0.001, and 34.0% (23.5, 34.0), P = 0.001, in absolute values versus baseline, respectively, and after restoration of flow rapidly improved. Thus, for brain saturations both ECMO regimens reached values exceeding 80% of baseline and similar ranges apply for IABP assistance. Comparably similar values were detected also for peripheral saturations. None of the comparisons for brain or peripheral saturations differed significantly.

Immediatelly after induction of VF and cardiac arrest, the CoPP decreased from a baseline of 84 mmHg (69.7, 97.0) to 15 mmHg (9.5, 20.8). Of note, animals randomized to FS starting cohort, tended to have nonsignificantly lower pre-arrest CoPP. The first CoPP value on ECMO was 34 mmHg (25.8, 45.5) and later during the protocol gradually rose. This phenomenon of a gradual rise with time was detected in almost all animals except for two of them, see Figure 3 and Additional file 4 and was present both in the FS and FF arms of the protocol. A more pronounced increase in CoPP values was detected in the FF ECMO arm (see Figure 4). Table 3 demonstrates the actual progress of CoPP, MAP and CVP values during the experiment and their respective differences according to a randomization assignment. Baseline pulmonary capillary wedge pressure and cardiac output values are also shown to demonstrate comparable baseline values in respect to treatment arms. Statistical evaluations comparing baseline CoPP values with values reached before CPR and ECMO onset values with those before CPR values are outlined in Figures 3 and 4. The final CoPP value at the end of the protocol, that is, after ECMO-treated cardiac arrest of more than two hours on average, reached almost the baseline value. Animals started on FF ECMO completed the protocol with identical CoPP values as at the baseline, with a steep rise in CoPP during the FF ECMO part of the protocol. The difference of initial CoPP ECMO values between the FF versus FS treatment arms was 10 mmHg (41 minus 31), P = 0.193, increased to a significant difference of 42.0 mmHg (77.0 minus 35.0) at the time of ECMO switch, P = 0.045, and remained significantly different before CPR, that is, 40.5 mmHg (86.0 minus 45.5 mmHg), P = 0.008. Animals starting on FS ECMO had a similar rise in CoPP during the later phase of the protocol, after a switch to FF ECMO. For more statistical differences, please see Figure 4.

Lactate levels measured in the coronary artery ostium and in the coronary sinus are outlined in the upper part of Figure 5. Interestingly, we have detected significantly higher levels of lactate in animals started on FS ECMO following cardiac arrest in all post arrest periods (P = 0.016 and P = 0.035 for differences in arterial and coronary sinus lactate levels, respectively). Moreover, after a steep increase of lactate levels during cardiac arrest and the first ECMO period, a plateau level and then a slight decrease could be detected, which was more prominent for the FF ECMO approach. The lower part of Figure 5 demonstrates O₂ extractions outlined similarly as lactate levels. After a steep increase in O₂ extraction during cardiac arrest a very deep decline of O₂ extraction related to ECMO initiation is apparent and these low extractions remain further with no statistically significant differences between the respective ECMO arms (P for difference = 0.547).

Resuscitability of the animals was high. Out of 11 animals, we gained 5 minutes ROSC in 8 animals (73%). 60 minutes ROSC was achieved in 8 (73%) animals (Additional file 5); however, see animal #3, in whom initial hypotension (5 minutes ROSC assessment with mean arterial pressure of 58 mmHg) improved markedly to fullfil criteria for 60-minute ROSC. However, the ECMO support of > 50 mL/kg/min was prolonged to keep perfusion pressure in 5 of 11 animals (#3, 4, 6, 8, 11). In the rest of them, ECMO was weaned quickly after initial CPR. Two animals could not be resuscitated to reach ROSC, one developed pulseless electrical activity after the first defibrillation (animal #1), the other suffered refractory VF despite six defibrillations (animal #2). Interestingly, these animals with unsuccessful CPR had steep decreases in coronary perfusion pressure during the ECMO phases, as described above, see Additional file 4.

ECMO and IABP have been implanted prior to induction of cardiac arrest, which is not the „real" clinical scenario model. However, urgent ECMO implantation (especially the subclavian approach) during cardiac arrest is a complex invasive procedure that would probably interfere with the primary aim of this study, that is, proper and accurate determination of coronary and carotid blood flow velocities following the cardiac arrest period. Blood flow measurements were not obtained directly using a flow probe, which, however, would need an open chest model. Instead, we assumed that Doppler wire acquired APV offers a good linear relation to blood flow in proximal LCx segments [19] and thus can be used as a surrogate marker of blood flow, mainly for detection of flow velocity changes assuming nonsignificant changes in a vessel diameter [20]. We controlled for potential confounding factors influencing vessel diameter like pCO₂, intravascular volume, level of vasocontriction and ECMO flow. Obtained APV values are also potentially influenced by proper wire positioning. To overcome this limitation, we insisted on a very careful wire handling to minimize position changes during the course of the study protocol.

In our cardiac arrest period, the "no-flow" phase was substituted entirely by a "low-flow" phase due to ECMO basal flow. However, basal flow was very low (targeted to 5 to 7, never over 10 mL/kg/minute) and monitored parameters (regional cerebral saturation, lactate, O₂ extraction) demonstrated profound changes corresponding to severe organ hypoperfusion. From the technical reasons, we also had to omit the mechanical chest compressions during the initial „low-flow" cardiac arrest, because Doppler wire technology is extremely sensitive to movements and any even very delicate changes in position, that is, mechanical compressions, would result in losing the proper position to catch and exactly record the post arrest flows.

For brain oxygenation, we used NIRS technology to monitor regional cerebral saturation with its inherent limitations; that is, not a global cerebral oxygenation assessment, not a deep tissue penetration, only relative values in one subject and their change in time can be used for evaluation. However, this is a very easy, noninvasive tool for detection of even minor changes in regional brain saturation and its use has been validated both in a pig CPR model [48] and for prognostic assessment in OHCA survivors on admission to hospital [49].

Different arterial cannula size used for FS (15 F) and FF (17 F) approach because of the vessel size is another limitation; however, target ECMO flow of 100 mL/kg/minute was achieved and did not differ in both approaches. And finally, our biomodel was represented by a breed that has been validated for simulation of human cardiac arrest and resuscitation [48, 50]. Nonetheless, pig anatomy differs from human and the presence of bicarotid trunk instead of two separate carotid arteries and the angle under which subclavian artery leaves the aorta might possibly influence our results.