Extracorporeal membrane oxygenation (ECMO) has gained renewed interest in the treatment of respiratory failure since the advent of the modern polymethylpentene membranes. Limited information exists, however, on the performance of these membranes in terms of gas transfers during multiple organ failure (MOF). We investigated determinants of oxygen and carbon dioxide transfer as well as biochemical alterations after the circulation of blood through the circuit in a pig model under ECMO support before and after induction of MOF. A predefined sequence of blood and sweep flows was tested before and after the induction of MOF with fecal peritonitis and saline lavage lung injury. In the multivariate analysis, oxygen transfer had a positive association with blood flow (slope = 66, P<0.001) and a negative association with pre-membrane PaCO2 (slope = −0.96, P = 0.001) and SatO2 (slope = −1.7, P<0.001). Carbon dioxide transfer had a positive association with blood flow (slope = 17, P<0.001), gas flow (slope = 33, P<0.001), pre-membrane PaCO2 (slope = 1.2, P<0.001) and a negative association with the hemoglobin (slope = −3.478, P = 0.042). We found an increase in pH in the baseline from 7.50[7.46,7.54] to 7.60[7.55,7.65] (P<0.001), and during the MOF from 7.19[6.92,7.32] to 7.41[7.13,7.5] (P<0.001). Likewise, the PCO2 fell in the baseline from 35 [32,39] to 25 [22,27] mmHg (P<0.001), and during the MOF from 59 [47,91] to 34 [28,45] mmHg (P<0.001). In conclusion, both oxygen and carbon dioxide transfers were significantly determined by blood flow. Oxygen transfer was modulated by the pre-membrane SatO2 and CO2, while carbon dioxide transfer was affected by the gas flow, pre-membrane CO2 and hemoglobin.
Citation: Park M, Costa ELV, Maciel AT, Silva DPe, Friedrich N, Barbosa EVS, et al. (2013) Determinants of Oxygen and Carbon Dioxide Transfer during Extracorporeal Membrane Oxygenation in an Experimental Model of Multiple Organ Dysfunction Syndrome. PLoS ONE 8(1): e54954. https://doi.org/10.1371/journal.pone.0054954
Editor: Lynette Kay Rogers, The Ohio State Unversity, United States of America
Received: August 17, 2012; Accepted: December 18, 2012; Published: January 29, 2013
Copyright: © 2013 Park et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: The authors have no support or funding to report.
Competing interests: The authors received a donation of PLS systems from MAQUET cardiovascular from Brazil, and Marcelo Park, Luciano Cesar Pontes Azevedo, Eduardo Leite Vieira Costa and Guilherme Pinto Paula Schettino were lecturers from three Brazilian courses of mechanical ventilation sponsored for MAQUET cardiovascular from Brazil. However, this does not alter the authors’ adherence to all the PLOS ONE policies on sharing data and material.
Extracorporeal membrane oxygenation (ECMO) has been successfully used to support patients with severe acute respiratory failure associated with refractory hypoxemia or uncompensated hypercapnia. – The choice of the ECMO settings depends on whether oxygenation or CO2 removal is the main indication of extracorporeal support. The concept that blood flow affects the transfer of oxygen and CO2, and sweep (gas) flow, that of CO2  is supported by evidence in healthy animals using a spiral coiled rubber silicone membrane.  There is, however, limited information about the main determinants of oxygen and carbon dioxide transfer during ECMO using the modern polymethylpentene membrane in subjects with multiple organ failure (MOF).  Many clinical decisions at the bedside are based on accumulated experience or on simple physiological assumptions and have not been formally tested. For example, in one study,  hemoglobin was empirically maintained at 14 g/dL with the aim of optimizing oxygen transfer. Many other variables readily available at the bedside, such as body temperature, blood osmolality, and baseline blood gases could also alone or in combination influence ECMO efficiency.;  The knowledge on which of these variables has the greatest impact on oxygenation and CO2 removal would be important to improve ECMO delivery.
During the passage of blood through the ECMO circuit, there are profound modifications of the partial pressure of oxygen (PO2) and of carbon dioxide (PCO2), and consequently pH.  These modifications affect the ionization of proteins  and their binding capacity ;  and cause electrolytes shifts between the intra and extra-cellular media or between intra and extra-vascular media.  These electrolytes alterations, although negligible in physiological conditions, may be unpredictable in non-physiological pH ranges.  Some of the ECMO-associated electrolyte disturbances can be clinically significant. For example, they can be associated with hemodynamic depression  and can affect the electrolyte clearance of renal replacement therapy when the blood is drained from the post-membrane ECMO circuit.  An adequate characterization of these post membrane alterations is of potential clinical importance and has not been carried out so far.
Based on basic physiological principles, we hypothesized that the oxygen transfer would be governed by the blood flow, hemoglobin concentration, blood temperature, and by pre-membrane oxygen saturation, while the CO2 transfer would be governed by temperature, gas flow, and by pH. Furthermore, we presumed that the passage of blood through the ECMO circuit would be associated with significant electrolyte disturbances. To test these hypotheses, we measured gas transfers as well as electrolyte concentrations pre and after membrane in animals supported with venous-venous ECMO, with assorted blood and sweep flows, before and after the induction of lung injury and sepsis. Our objective was to assess the determinants of oxygen and carbon dioxide transfer, and biochemical disturbances during the passage of blood through the polymethylpentene extracorporeal membrane applied to normal and MOF subjects.
This study had approval from the Institutional Animal Research Ethics Committee from the Hospital Sírio Libanês in São Paulo – Brazil and was performed according to National Institutes of Health guidelines for the use of experimental animals. Instrumentation, surgical preparation, pulmonary injury, and induction of sepsis were performed as previously described. –.
Instrumentation and Surgical Preparation
The room temperature was set at 24° Celsius. Five domestic female Agroceres pigs (80 [79,81] kg) were anesthetized with thionembutal (10 mg/kg, Tiopental, Abbott, Brazil) and pancuronium bromide (0.1 mg/kg, Pavulon, AKZO Nobel, Brazil) and connected to a mechanical ventilator (Evita XL Dräger, Dräger, Luebeck, Germany) with the following parameters: tidal volume 8 mL/kg, end-expiratory pressure 5 cm H2O, FiO2 initially set at 100% and subsequently adjusted to maintain arterial saturation between 94–96%, and respiratory rate titrated to maintain PaCO2 between 35 and 45 mm Hg or an end-tidal CO2 (NICO, Dixtal Biomedica Ind. Com, Sao Paulo, Brazil) between 30 and 40 mmHg. The electrocardiogram, heart rate, oxygen saturation, and pressures of the animals were monitored with a multiparametric monitor (Infinity Delta XL, Dräger, Luebeck, Germany). Anesthesia was maintained during the study with midazolam (1–5 mg.kg−1.h−1) and fentanyl (5–10 mcg.kg−1.h−1) and muscular relaxation with pancuronium bromide (0.2 mg.kg−1.h−1). The adequate depth of anesthesia during the surgical period was evaluated with maintenance of physiological variables (heart rate and arterial pressure) and absence of reflexes (corneal and hind limb flexion response), as well as unresponsiveness to stimuli during manipulation. Supplementary boluses of 3–5 mcg/kg fentanyl and 0.1–0.5 mg/kg midazolam were administered as necessary.
The left external jugular vein was cannulated (guided by ultrasonography) to introduce a pulmonary artery catheter, and the right external jugular vein, to introduce a 25-cm ECMO return canulae (Edwards Lifesciences, Irvine, CA, USA). The right femoral vein was punctured for the insertion of a 55-cm ECMO drainage canulae (Edwards Lifesciences, Irvine, CA, USA), which was positioned close to the right atrium with the aid of trans-hepatic ultrasonographic visualization. Only the guidewires were inserted until the first baseline measurements after the stabilization period, when they were replaced by the ECMO catheters. After the insertion of the guidewires, an infusion of 1000 IU per hour of heparin was started. A central venous catheter and an invasive arterial blood pressure catheter were placed in the left femoral vein and artery, respectively.
Through a midline laparotomy, a cistostomy was performed, and a bladder catheter inserted. A 2-cm incision was performed in the descending colon, and 1.0 g/kg of fecal contents was removed and stored. The intestinal incision was sutured; two large-bore catheters were inserted in each flank of the animal, and the laparotomy was closed. During the surgical interventions, 15 mL.kg−1.h−1 of lactated Ringer’s were continuously infused, and boluses of 250 mL were used to maintain a systemic mean arterial blood pressure (ABPm) of 65 mmHg or greater, central venous pressure (CVP) of 8 mmHg or greater, and SvO2 greater than 65% until the end of instrumentation.
Stabilization and Support of the Animals
After the end of the instrumentation, the animals were allowed to stabilize for 1 h. A continuous infusion of 3 mL.kg−1.h−1 of lactated Ringer’s was started at the beginning of the stabilization period and kept throughout the entire experiment. Blood glucose was monitored at least hourly. If the blood glucose was lower than 60 mg/dL, a solution of 40 mL of 50% glucose was infused.
When the animals became hypotensive (ABPm<65 mmHg), a bolus of 500 mL of lactated Ringer’s was infused. If the ABPm failed to rise above 65 mmHg after the bolus, an infusion of norepinephrine 0.1 mcg.kg−1.min−1 (Norepine, Opem Pharmaceuticals, São Paulo, Brazil) was started and titrated to an ABPm ≥ 65 and <80 mmHg.
ECMO Priming, Starting, and Maintenance
The ECMO system (Permanent life support system - PLS, Jostra – Quadrox D, Maquet Cardiopulmonary, Hirrlingen, Germany) was primed with a 37° Celsius normal saline solution and connected to a centrifugal pump (Rotaflow, Jostra, Maquet Cardiopulmonary, Hirrlingen, Germany). With the circuit filled, 1000 IU of heparin were injected in the circulating fluid. The anticoagulation was monitored with the measurement of the activated coagulation time (ACT) at baseline and every 6 hours. The infusion of heparin was titrated to keep the ACT 1.5–2.5 times the first baseline ACT value. The CO2 partial pressure of the air exhaled from the respiratory membrane was measured with an EtCO2 probe Tonocap® (Datex -General Electric Healthcare, USA).
The PLS uses a polymethylpentene membrane; the tubes are coated with a bioactive and biopassive system (Bioline, Maquet Cardiopulmonary, Hirrlingen, Germany).  Two luer-locks were connected respectively in the pre and post membrane ports, in order to allow for the measurement of pressures and for the collection of blood samples. Pressure lines were connected to ports in the drainage tube (before the centrifugal pump), pre and post ECMO membrane; pressure measurements were performed in real time with a multiparametric monitor (Dx 2020, Dixtal Biomedica Ind. Com, Sao Paulo, Brazil).
During the extracorporeal circulation, if the system began to vibrate significantly or cavitate, or if the blood flow decreased despite unchanged rotations per minute in the pump, a standardized sequence of procedures was done: first, the position of the animal was gently modified with lateralization of the animal, and if necessary followed by a semi-recumbent positioning; second, the PEEP level was raised in 1 or 2 cmH2O. If the PEEP was higher than 6 cmH2O, a trial to lower the PEEP by 1 or 2 cmH2O could also be performed; third, boluses of 250 mL of lactated Ringer’s were tried.
The following data were collected: heart rate (HR), mean arterial blood pressure (ABPm), central venous pressure (CVP), mean pulmonary artery pressure (PAPm), pulmonary artery occluded pressure (PAOP), cardiac output (CO), core temperature, peripheral oxygen saturation, end-tidal CO2 (EtCO2), and mixed venous oxygen saturation (SvO2), PEEP, FiO2, auto-PEEP measured with 4 seconds of expiratory pause, plateau pressure with 2 seconds of inspiratory pause, and peak pressure. Blood samples from the pulmonary and femoral arteries were collected and analyzed in a standard radiometer ABL 600 (Radiometer, Copenhagen, Denmark). The sample from the femoral artery was used for the biochemical analysis.
We collected the baseline data after 30 min of ECMO at a blood flow of 1.5 L/min and with the gas (sweep) flow turned off, i.e. without gas exchange. The formulas for calculations of respiratory, hemodynamic, and ECMO parameters are available in the electronic supplement that accompanies this manuscript. The standard formulas used to the calculus ; ; ;  are shown in the Text S1– Online supplement.
Varying Blood and Sweep Flow
A sequence of blood and sweep flows was chosen based on the usual needs at bedside. In this way, a blood flow of 1500 and 3000 mL are commonly used in ECMO support, and a blood flow of 500 mL can be reached during interventional lung assist support. The sweep flows were chosen in order to allow measurements of gas transfer in a blood/sweep flow ratio of 2∶1, 1∶1 and 1∶2.
After the baseline measurements, the blood flow was set to 5000 mL/min and the sweep flow, to 5.0 L/min. After 10 minutes, the blood flow was reduced to 1500 mL/min, and the sweep flow was set at two-minute intervals to 1.5, 1.0, and 3.0 L/min. The blood flow was then increased to 3000 mL/min, and the sweep flow was set to 1.5, 3.0, and 6.0 L/min in two-minute steps. Finally, the blood flow was reduced to 500 mL/min and the sweep flow was set using 0.25, 1.0, and 0.5 L/min at two-minute intervals. At the end of each sweep flow step, blood samples pre and post-membrane were collected, as well as the EtCO2 value from the exhalation port of the respiratory membrane.
After this first sequence of data collection, we induced lung injury through surfactant depletion (lung lavage with aliquots of 1 liter of normal saline at 37°Celsius until the P/F ratio was <50) and we induced sepsis due to fecal peritonitis. The peritonitis was induced through the injection of 1 g/kg of feces into the peritoneal cavity. Antimicrobial treatment with 15 mg/kg of amikacin and 500 mg of metronidazol was administered four hours after the peritonitis induction. This interval was used because in previous experiments this period was associated with higher plasmatic interleukin-6 concentrations.  Amikacin and metronidazole were used due to the adequacy in peritoneal-fecal septic shock in previous studies. ,  The PEEP used was titrated after a stepwise alveolar recruitment maneuver, in steps of 2 cmH2O down from 25 cmH2O, looking for the best dynamic compliance. The PEEP was set 2 cmH2O higher than the PEEP corresponding to the best compliance. After the induction of peritonitis, a period of twelve hours was awaited for the MOF installation, based on previous experiments.;  Hallmarks of MOF in the model are persistent hypotension (ABPm<65 mmHg for 30 minutes), metabolic, renal, and respiratory dysfunctions. Table S1 shows the characterization of the organs failure during the twelve hours evolution and immediately before the second sequence of data collection. After MOF installation, the same sequence of data collection described above was repeated (except for the baseline with and without gas flow). The timeline of the study is shown in the figure 1.
Phase 1 is the period of anesthesia, instrumentation and stabilization of animals; Phase 2 is the period of data collection to analyze gas transfers in the normal animals; and Phase 3 is the phase of equilibrium analysis *, lung lavage, peritonitis induction, and multiple organ failure installation. At the end of phase 3 the data collection described in the phase 2 was repeated, now in the animals with multiple organ failure installed. Gray grounds show the periods of data collection to the current manuscript. *The equilibrium data collection and analysis are the focus of other manuscript.
Classically CO2 transfer is measured by computing the CO2 exhaled from the ECMO device. This approach avoids measuring blood CO2 transfer through the trans-membrane CO2 gradient, which can suffer from inaccuracies in the calculation of blood CO2 contents, especially with major pH and pCO2 variations during the passage of blood through the respiratory membrane.  To explore these two different methods, we measured the correlation and agreement of both techniques. For all other analyses, we used the gold standard CO2 transfer computed using the CO2 exhaled from the respiratory membrane.
Data were predominantly non-normal as assessed by the Shapiro-Wilk goodness-of-fit model and are shown as median and percentiles 25th and 75th. Non-paired data were analyzed with the Mann-Whitney’s or Kruskal-Wallis` tests as appropriate; paired data were analyzed with Wilcoxon`s and Friedman’s test. Post-hoc analyses were performed using the Tukey`s “Honestly Significantly Different” (HSD) test. Correlation and agreement analyses were done between the CO2 transfer measured through the blood CO2 gradient pre- and post-membrane and the CO2 transfer measured through the CO2 exhaled from the respiratory membrane with the correlation scatter plot, the Pearson correlation method, and the Bland – Altman diagram. Multivariate analyses were performed using a backward elimination mixed generalized model with the animal as random factors in order to account for the within-subject correlation among repeated observations. The Markov chain Monte Carlo procedure with 10,000 simulations to reach the equilibrium of distributions was used to retrieve a fixed probability of each resulting independent variable from the mixed generalized model after the backward elimination. Collinearity among the independent variables was tested with the Spearman’s test of correlations in a matrix including all variables tested. Variables with the r coefficient >0.85 were further tested for multicollinearity with the variance inflation factor (VIF). Variables with a VIF<2.5 were considered appropriate to the analyses. The pseudo-R2 was calculated for each model in order to show their goodness-of-fit; the calculation was performed with the squared ratio between the Spearman correlation between the fitted values of the model and the original values retrieved from the experiment. The R free source statistical package and comprehensive-R archive network (CRAN)-specific libraries were used to build the graphics and to do all the statistical analyses .
Five animals were studied. Four animals used a 20-French catheter and one animal used a 21-French catheter to drain the blood to the ECMO device. Three animals used 21-French return catheter from the ECMO system, and two used 20-French catheter. Data on the baseline hemodynamics, respiratory variables, and support measures collected at baseline and just before the second sequence of data collection are described in table 1. The total number of lung lavages needed to achieve lung injury was 8, 12, 16, 4, and 6, to reach a PaO2 of 54, 50, 50, 46 and 47 mmHg in each animal respectively. The animals were ventilated with a PEEP = 10 cm H2O and a FiO2 = 1 at this time. The second sequence of data collection, after the induction of multiorgan failure, occurred 1420 [1384,1508] minutes after the first sequence of data collection and 735 [720,740] minutes after the peritonitis induction. Table S1 shows the characterization of MOF at the time of the second sequence of data collection, and at the worst organ function measured within the 735 [720,740] minutes between peritonitis induction and the second sequence of data collection. The fluid balance between the periods of data collection was positive in 8386 [6734,14534] mL of lactated Ringer’s solution. The ACT controls 6, 12, and 18h after the first baseline measurement were 2.0 [2.0,2.6], 2.2 [1.7,2.6] and 1.8 [1.7,2.0] times the control (first baseline) respectively.
Figure S1 shows the correlation and agreement between the CO2 transfer measured through the CO2 exhaled from the respiratory membrane and measured through the trans-membrane gradient of CO2. In spite of the good correlation and low bias, the distribution of differences between the upper and lower limits of the Bland-Altman plot precludes the use of one technique instead the other. As already posed, we used the classical CO2 exhalation technique.
Determinants of Oxygen and Carbon Dioxide Transfers
Two multivariate models were built (Table 2). The first, to assess the association between oxygen transfer as the dependent variable and blood flow, oxygen saturation, PaCO2, temperature, base excess, pH, gas flow, and hemoglobin as independent variables. The second, to assess the association between carbon dioxide transfer and blood flow, oxygen saturation, PaCO2, temperature, sweep flow, and hemoglobin. In both models, the laboratory data were obtained from the pre-membrane port. In the multivariate analysis, oxygen transfer was associated with ECMO blood flow, pre-membrane SatO2, and pre-membrane PaCO2. The CO2 transfer, on the other hand, was associated with gas flow, pre-membrane PaCO2, blood flow and negatively with hemoglobin. The graph correlating the variations in the main variables associated with gas transfers in the multivariate analysis and gas transfers per se are shown in the spider plot presented in the figure 2. The marginal-model plots extracted from the regression model are shown in figure 3, in order to facilitate the understanding of the isolated effect of each variable.
All data were collected from the animals, with and without multiple organ failure. The input variables are shown in the legend.
The default variables were: sweep flow = 2 L/minute, initial pre-membrane SatO2 = 30%, initial pre-membrane PCO2 = 30 mmHg and initial hemoglobin = 8.0 g/dL. The marginal model is derived from the multivariate mixed model. Panel A shows the CO2 transference resultant from sweep flow variation. Panel B shows the CO2 transference resultant from hemoglobin variation. Panel C shows the CO2 transference according to the initial PCO2 variation. Panel D shows the O2 transference resultant from ECMO blood flow variation. Panel E shows the O2 transference resultant from initial PCO2 variation. Panel F shows the O2 transference according to the initial pre-membrane SatO2 value.
Laboratory Results Pre and Post Membrane
Table 3 shows the laboratory results before and after the passage of blood through the membrane both in health and after the induction of multiple organ failure. We found an increase in the pH in the post-membrane blood, a slight reduction in the ionized calcium value and a rise in the chloride concentration after the passage of blood through the membrane. The elevation in the pH was more accentuated during MOF. The effects of different ECMO settings on the laboratory results are shown in tables 4 and 5. Additionally, these tables explore the effect of multiorgan failure on the laboratory results. The pH variation is of note, mainly in MOF measurements, associated with a higher pre-membrane PCO2 and lower pre-membrane oxygen saturation. The higher lactate levels in MOF measurements also deserve attention. The potassium levels were higher during MOF; however the effect of membrane passage on the electrolytes were the same in normal and MOF measurements.
Our main findings were 1) both oxygen and carbon dioxide transfers were significantly determined by blood flow; 2) the oxygen transfer was also affected by the pre-membrane oxygen saturation, pre-membrane CO2 and the carbon dioxide transfer, by the gas flow, hemoglobin level, and the pre-membrane CO2; 3) there were no concerning electrolyte disturbances associated with the passage of blood through the ECMO system. The slight electrolyte modifications seen are not sufficient to trigger specific interventions at the bedside, suggesting statistical but not clinical significance.
The Animal Model
This model was created in order to keep many of the multiorgan dysfunction characteristics seen in intensive care units at the bedside. The cardiovascular, renal, and respiratory functions were moderately or severely impaired (Table S1).
The respiratory static compliance reduction could be partially explained by the PEEP elevation leading to overdistention. However, besides the fall in the static compliance, we also observed a dramatic decrease in oxygenation (P/F ratio of 50 [47,51] mmHg), caused by the elevation of the pulmonary shunt to 45 [38,69]% at the MOF baseline, indicating that we obtained a significant amount of lung collapse. Low compliance associated with lung collapse is an important part of the characterization of ARDS (the baby lung) even if, alike in patients, the component related to the disease severity cannot be separated from the application of higher PEEP levels.
Oxygenation and Carbon Dioxide Removal
In the multivariate analysis (Table 2), the oxygen transfer was associated with ECMO blood flow, pre-membrane SatO2, and pre-membrane PaCO2. The association of O2 transfer with blood flow is in line with the current understanding of the gas exchange during ECMO.  The lower pre-membrane oxygen content together with the maximized post-membrane oxygen content would explain the higher O2 transfers with lower pre-membrane SatO2. On the other hand, our finding of higher O2 transfers attained at lower pre-membrane PaCO2 values cannot be ascribed to an increase in the availability of oxygen binding sites in the pre-membrane blood, because low PaCO2 values are associated with higher hemoglobin affinity for oxygen and a higher pH.  Indeed, we were unable to bring forth a physiological explanation for that finding. In a speculative way, once the low PaCO2 is associated with a higher oxygen-hemoglobin affinity, a low pre-membrane PaCO2 offers a higher contact surface between blood and the lung membrane in an environment with a higher oxygen-hemoglobin affinity, resulting in a higher oxygen transfer. We cannot exclude, however, that there was some residual confounding between ECMO blood flow and PaCO2 even after the multivariate adjustment, considering that lower values of pre-membrane PaCO2 were associated with higher ECMO blood flows (Table 3).
The CO2 transfer was associated with gas flow, pre-membrane PaCO2, blood flow and negatively with hemoglobin (Table 5). The strong association between CO2 transfer and gas flow is well described and frequently used at the bedside.  We also found an association previously described by Kolobow et al in silicone rubber membranes,  between high pre-membrane PaCO2 values and high CO2 transfers. Of note, the increase in the CO2 transfer caused by a 10 mmHg increase in the PaCO2 was clinically significant and comparable to an increase by 0.5 L/min in the flow of gas. This finding could be of value at the bedside supporting the use of permissive hypercapnia to help optimize the ECMO CO2 removal. Another interesting finding was the positive correlation between blood flow and CO2 transfer. The effect of blood flow on the CO2 transfer was roughly 1.9 times that of the gas flow (Table 2), suggesting, for example, that the pumpless (arteriovenous) systems may have lower efficiency not only in terms of oxygenation, but also regarding CO2 removal. The negative association between the CO2 transfer and hemoglobin can possibly be explained by a higher CO2 content in the form of bicarbonate in the presence of low hemoglobin levels, which could facilitate in an acid pH in the formation of CO2, which is easily eliminated through the membrane. In this scenario, the Bohr effect promoting CO2 liberation from the hemoglobin during the membrane oxygenation is probably of reduced significance.
In agreement with the preceding multivariate analyses, when the data were stratified among the three blood flows used, the O2 transfer increased six times from the blood flow = 0.5 L/min to blood flow = 3.0 L/min, and the CO2 transfer increased two and a half times (Table 4). The elevation of the gas flow from 0.5 to 2 times the blood flow caused a significant augmentation in the pre-post membrane gradient of CO2, consequently increasing the CO2 transfer. In contrast, the O2 transmembrane gradient and the O2 transfer were the same in all sweep/blood flow ratios, suggesting no effect of the gas flow on the oxygenation within the range studied (Table 5). The oxygen transfer, with the ECMO blood flow set in 500 mL/min (Table 4), was lower in the pre-multiorgan failure measurements than in the multiorgan dysfunction phase. This finding could be explained by the strikingly low pre-membrane oxygen saturation, associated with a relatively low pre-membrane PCO2 during MOF. The multiorgan failure CO2 transfers were consistently higher in the multiorgan measurements than in the normal measurements (Table 4 and Table 5), possibly due to higher pre-membrane PCO2 in the MOF measurements.
The pre-membrane PCO2 and oxygen saturation are cardiac output- dependent variables. In this way, one can conclude that the cardiac output is a potential gas transfer modulator in patients ECMO supported. This fact deserves more investigation.
Laboratory Results Pre and Post Membrane
During the passage of blood through the oxygenation membrane, there was an increase in pH, a slight reduction in the ionized calcium, and an increase in the chloride, more accentuated after the induction of the MOF (Table 3; Table 4 and table 5). The Gibbs-Donnan effect,;  secondary to the ionization of proteins due to the pH variation could explain, at least in part, the chloride and calcium variations through intracellular or extra-vascular shifts.  Although statistically significant, the slight reduction in the calcium concentration was not clinically relevant, different from the previously described hemodynamic impairment associated with hypocalcemia in ECMO supported children.  Likewise, the changes in chloride concentrations were of minor clinical significance. These findings diminish the hypothesis that the ECMO-induced modifications of the pH  could cause significant electrolytes shifts between the intra and extra-cellular media.; .
This study has several limitations: 1) higher O2 transfer in the ECMO system does not necessary mean higher oxygen availability to the tissues;  2) the low number of animals used could lead to type II errors, attenuating the validity of our findings of lack of associations. Of note, such limitation would not affect our positive findings; 3) in this manuscript; we focused only on the gas transfer across the respiratory membrane. The analysis of the interaction between ECMO blood flow and cardiac output at different natural lung shunt will be explored elsewhere.
In summary, we confirmed the common knowledge that blood and gas flows affect oxygen transfer and carbon dioxide removal. We also found that the oxygen transfer was negatively associated with the pre-membrane oxygen saturation and carbon dioxide partial pressure. The carbon dioxide transfer was positively associated with the pre-membrane carbon dioxide partial pressure and blood flow, and negatively associated with the hemoglobin level. There was a clinically significant transmembrane elevation of chloride, and pH, however only the former was associated with blood and gas flow variations.
Correlation and agreement between the CO2 transfer in the lung membrane, measured through the CO2 exhaled from the membrane and through the blood CO2 content fall during membrane passage. Panel A shows the correlation and Panel B shows the agreement.
Characterization of multiorgan dysfunction.
We would like to thank the Hospital das Clínicas de São Paulo and Hospital Sírio-Libanês ECMO group: Leandro Utino Taniguchi, Raquel Oliveira Nardi, Michelle de Nardi Ignácio, Cláudio Cerqueira Machtans, Wellington Alves Neves, Pedro Mendes, Marcelo Brito Passos Amato and Carlos Roberto Ribeiro Carvalho.
The study was done in the Research and Education Institute of Hospital Sírio Libanês in São Paulo – Brazil.
Conceived and designed the experiments: MP ELVC ATM GPPS LCPA. Performed the experiments: MP ELVC ATM DPeS NF EVSB ASH GPPS LCPA. Analyzed the data: MP ELVC LCPA. Wrote the paper: MP ELVC LCPA GPPS ATM.
- 1. Peek GJ, Mugford M, Tiruvoipati R, Wilson A, Allen E, et al. (2009) Efficacy and economic assessment of conventional ventilatory support versus extracorporeal membrane oxygenation for severe adult respiratory failure (CESAR): a multicentre randomised controlled trial. Lancet 374: 1351–1363.
- 2. Davies A, Jones D, Bailey M, Beca J, Bellomo R, et al. (2009) Extracorporeal Membrane Oxygenation for 2009 Influenza A(H1N1) Acute Respiratory Distress Syndrome. JAMA 302: 1888–1895.
- 3. Noah MA, Peek GJ, Finney SJ, Griffiths MJ, Harrison DA, et al. (2011) Referral to an Extracorporeal Membrane Oxygenation Center and Mortality Among Patients With Severe 2009 Influenza A(H1N1). JAMA.
- 4. Patroniti N, Zangrillo A, Pappalardo F, Peris A, Cianchi G, et al. (2011) The Italian ECMO network experience during the 2009 influenza A(H1N1) pandemic: preparation for severe respiratory emergency outbreaks. Intensive Care Med 37: 1447–1457.
- 5. Sidebotham D, McGeorge A, McGuinness S, Edwards M, Willcox T, et al. (2009) Extracorporeal Membrane Oxygenation for Treating Severe Cardiac and Respiratory Failure in Adults: Part 2-Technical Considerations. J Cardiothorac Vasc Anesth.
- 6. Kolobow T, Gattinoni L, Tomlinson T, White D, Pierce J, et al. (1977) The carbon dioxide membrane lung (CDML): a new concept. Trans Am Soc Artif Intern Organs 23: 17–21.
- 7. Chauhan S, Subin S (2011) Extracorporeal membrane oxygenation, an anesthesiologist’s perspective: physiology and principles. Part 1. Ann Card Anaesth 14: 218–229.
- 8. Figge J, Rossing TH, Fencl V (1991) The role of serum proteins in acid-base equilibria. J Lab Clin Med 117: 453–467.
- 9. Morgan TJ (2009) The Stewart approach–one clinician’s perspective. Clin Biochem Rev 30: 41–54.
- 10. Nguyen MK, Kurtz I (2006) Quantitative interrelationship between Gibbs-Donnan equilibrium, osmolality of body fluid compartments, and plasma water sodium concentration. J Appl Physiol 100: 1293–1300.
- 11. Meliones JN, Moler FW, Custer JR, Snyder SJ, Dekeon MK, et al. (1991) Hemodynamic instability after the initiation of extracorporeal membrane oxygenation: role of ionized calcium. Crit Care Med 19: 1247–1251.
- 12. Bosticardo GM (2010) The diffusion gradient between ionized calcium in dialysate and plasma water–corrected for Gibbs-Donnan factor–is the main driving force of net calcium balance during haemodialysis. Nephrol Dial Transplant 25: 3458–3459.
- 13. de Azevedo LC, Park M, Noritomi DT, Maciel AT, Brunialti MK, et al. (2007) Characterization of an animal model of severe sepsis associated with respiratory dysfunction. Clinics 62: 491–498.
- 14. Almeida J, Machado FS, Schettino GP, Park M, Azevedo LC (2010) Cardiopulmonary effects of matching positive end-expiratory pressure to abdominal pressure in concomitant abdominal hypertension and acute lung injury. J Trauma 69: 375–383.
- 15. Rosario AL, Park M, Brunialti MK, Mendes M, Rapozo M, et al. (2011) SvO2-Guided Resuscitation for Experimental Septic Shock: Effects of Fluid Infusion and Dobutamine on Hemodynamics, Inflammatory Response and Cardiovascular Oxidative Stress. Shock.
- 16. Zimmermann AK, Weber N, Aebert H, Ziemer G, Wendel HP (2007) Effect of biopassive and bioactive surface-coatings on the hemocompatibility of membrane oxygenators. J Biomed Mater Res B Appl Biomater 80: 433–439.
- 17. Siggaard-Andersen O (1977) The van Slyke equation. Scand J Clin Lab Invest Suppl 37: 15–20.
- 18. Douglas AR, Jones NL, Reed JW (1988) Calculation of whole blood CO2 content. J Appl Physiol 65: 473–477.
- 19. R Development Core Team (Viena - Austria) (2009) R: A language and environment for statistical computing. R Foundation for Statistical Computing.
- 20. Mulhausen RO (1970) The affinity of hemoglobin for oxygen. Circulation 42: 195–198.
- 21. Kellum JA, Bellomo R, Kramer DJ, Pinsky MR (1998) Etiology of metabolic acidosis during saline resuscitation in endotoxemia. Shock 9: 364–368.
- 22. Riggs TE, Shafer AW, Guenter CA (1973) Acute changes in oxyhemoglobin affinity. Effects on oxygen transport and utilization. J Clin Invest 52: 2660–2663.