Authors: Michael C. Müller, Sarah K. Wilke, Andrej Dobbermann, Sascha Kirsten, Martin Ruß, Steffen Weber-Carstens, Tobias Wollersheim
Categories: Adult Circulatory Support, ECLS, ECMO, dissolved oxygen, extracorporeal membrane oxygenation, gas exchange, intensive care, membrane lung, membrane lung performance, monitoring, oxygen transport, oxygen uptake, vvECMO
Source: Asaio Journal
When determining extracorporeal oxygen transfer (VMLO2) during venovenous extracorporeal membrane oxygenation (VV ECMO) dissolved oxygen is often considered to play a subordinate role due to its poor solubility in blood plasma. This study was designed to assess the impact of dissolved oxygen on systemic oxygenation in patients with acute respiratory distress syndrome (ARDS) on VV ECMO support by differentiating between dissolved and hemoglobin-bound extracorporeal oxygen transfer. We calculated both extracorporeal oxygen transfer based on blood gas analysis using the measuring energy expenditure in extracorporeal lung support patients (MEEP) protocol and measured oxygen uptake by the native lung with indirect calorimetry. Over 20% of VMLO2 and over 10% of overall oxygen uptake (VO2 total) were realized as dissolved oxygen. The transfer of dissolved oxygen mainly depended on ECMO blood flow (BFML). In patients with severely impaired lung function dissolved oxygen accounted for up to 28% of VO2 total. A clinically relevant amount of oxygen is transferred as physically dissolved fraction, which therefore needs to be considered when determining membrane lung function, manage ECMO settings or guiding the weaning procedure.
Keywords: extracorporeal membrane oxygenation, ECMO, vvECMO, ECLS, membrane lung, intensive care, gas exchange, dissolved oxygen, oxygen uptake, oxygen transport, monitoring, membrane lung performance
Oxygenation is an essential process to maintain vital organ functions. Inadequate systemic oxygen delivery leads to tissue hypoxemia and, depending on the severity of hypoxemia as well as tissue type, cellular dysfunction or even cell death.^1^ Under physiologic circumstances, more than 98% of the oxygen is transported by hemoglobin (Hb, g/dl) and only a small amount of oxygen is transported as dissolved gas by the blood.^2^
When the oxygenation capacity of the lungs decreases to an insufficient degree due to disease or trauma and cannot be maintained sufficiently with lung-protective mechanical ventilation, extracorporeal membrane oxygenation (ECMO) can be implemented as rescue intervention to sustain systemic oxygen delivery.^3^ In the absence of cardiac failure venovenous ECMO (VV ECMO) configuration is primarily used.^4^ In severe cases of acute respiratory distress syndrome (ARDS) oxygenation can solely depend on extracorporeal oxygen transfer. To ensure adequate oxygen delivery during ECMO therapy, membrane lung (ML) function and oxygen transport capacity need to be monitored closely.^5^
Therefore, devices that assess the gas exchange of the ML have been developed (e.g., Quantum Diagnostics System, Spectrum Medical, Cheltenham, England; Landing ECMO, Eurosets, Medolla, Italy), but the contribution of dissolved oxygen to extracorporeal oxygen uptake (VMLO2, ml/min) is neglected by manufacturers, most likely due to technical problems measuring gas partial pressures continuously as well as the minimal impact of dissolved oxygen for systemic oxygenation under physiologic conditions.^6,7^
Considering the established formula to calculate oxygen content of human blood including Hüfner’s constant und Bunsen solubility coefficient for oxygen, which seems conclusive at first.^8^
Assuming a steady state, an Hb concentration of 14 g/dl, oxygen partial pressure (PO2, mmHg) of 100 mmHg and oxygen saturation (SO2, %) of 100% this equates to an oxygen content (CbloodO2, ml/l) of 187.6 ml/l hemoglobin bound oxygen and only 3.1 ml/l of dissolved oxygen.
However, when 100% oxygen is used as sweep gas during VV ECMO support—a common practice in most ECMO centers—resulting supra-physiologically high PO2 values in postmembrane blood (PpostO2, mmHg) lead to a relevant increase of dissolved oxygen. The clinical relevance of dissolved oxygen for systemic oxygenation during VV ECMO therapy is discussed by in the literature despite the absence of studies investigating this problem.^9–11^
We hypothesize that the amount of dissolved oxygen has a clinically relevant impact on oxygenation during VV ECMO support. To test our hypothesis, we assessed extracorporeal oxygen transfer (VMLO2, ml/min), and calculated hemoglobin bound (VMLO2 Hb, ml/min) and dissolved oxygen transfer (VMLO2 dissolved, ml/min) via ECMO separately. Second, we asked which factors affect VMLO2 dissolved and if they differ from the determinants of VMLO2 and VMLO2 Hb.
This is a retrospective secondary analysis of a single-center, prospective, observational study. Between 2013 and 2021, we included 27 mechanically ventilated adult ARDS patients according to the Berlin definition who were admitted to a German ARDS referral center for VV ECMO support.^12^ Informed consent was obtained from the patients’ legal proxies before all investigations. This study was approved by the local ethics committee (Ethikkommission Charité - Universitätsmedizin Berlin EA1/293/13). The study protocol was registered under https://clinicaltrials.gov, ClinicalTrials.gov Identifier: NCT01992237. Earlier publications of this trial investigated the caloric needs of ECMO patients, as well as the feasibility and accuracy of a standalone device to measure gas exchange of the ECMO membrane in real-time.^13,14^
Indirect calorimetry (IC) was performed to assess the native lung function and determine caloric needs. Cosmed Quark RMR (COSMED, Rome, Italy) was used for IC. Setup and measurements were conducted according to the manufacturer’s manual after a steady state (e.g., no disconnection of mechanical ventilation, no endotracheal suction, steady breathing pattern) was maintained for at least 20 minutes. Measurement of oxygen uptake (VNLO2, ml/min) and carbon dioxide elimination by the native lung was continued for a minimum of 25 minutes.
We collected two additional blood samples for blood gas analysis (BGA) from the ECMO circuit before and after the ML during IC. Analysis was performed with Radiometer ABL 800 FLEX (Radiometer, Brønshøj, Denmark) blood gas analyzer. Indirect calorimetry and blood gas sampling was repeated each day, provided a change in ECMO settings, till ECMO weaning was completed or measurement was not possible for other reasons, for example, extubation, death, transfer to another hospital.
The calculations for extracorporeal gas transfer are derived from the previously published measuring energy expenditure in extracorporeal lung support patients (MEEP) approach, which aims at determining energy expenditure in ECMO patients.^14^ Arterial, pre-, and postmembrane blood oxygen content (CaO2, Cpre-MLO2, Cpost-MLO2, ml/l) was calculated with a model published by Dash et al.^15^ Fractions of blood oxygen content, hemoglobin bound (CaO2 Hb, Cpre-MLO2 Hb, Cpost-MLO2 Hb, ml/l), and dissolved oxygen (CaO2 dissolved, Cpre-MLO2 dissolved, Cpost-MLO2 dissolved, ml/l) were calculated separately. VMLO2 as well as VMLO2 Hb and VMLO2 dissolved were calculated applying the Fick principle according to the MEEP
The patients’ total oxygen uptake (VO2 total, ml/min) was calculated by adding VMLO2 and VNLO2.
The absolute amount of VMLO2 and consequently the absolute amounts of VMLO2 dissolved and VMLO2 Hb can vary between patients. Therefore, we calculated the percentage of VMLO2 dissolved in VMLO2 (VMLO2 dissolved pct, %), as well as the percentage of VMLO2 Hb in VMLO2 (VMLO2 Hb pct, %), for better comparison.
Oxygen transfer in VV ECMO depends on multiple factors, especially BFML and premembrane oxygen saturation (SO2 pre, %), as demonstrated by Park et al.^16^ and Schmidt et al.^17^ Furthermore, the oxygenation capacity deteriorates over time as a result of increasing extracorporeal shunt and dead space.^18,19^ We registered BFML and ECMO sweep gas flow (GFML), which are measured by the ECMO console, noted SO2pre, and Hb from BGAs to assess the dependency of VMLO2, VMLO2 dissolved, and VMLO2 Hb. PpostO2 is often used as surrogate parameter to evaluate ML function. To analyze whether VMLO2 is reflected by PpostO2, values were obtained from postmembrane BGA.
Data are shown as mean and standard deviation (±SD) or as declared. Box and whisker plots are used to display the distribution of data. Scatterplots with the best-fit line of a simple linear regression visualize the relationship between VMLO2, VMLO2 dissolved, VMLO2 Hb and the variables BFML, GFML, PpostO2, SO2pre, Hb, and VNLO2. The predictable proportion of variation in the dependent variable is expressed as R^2^. The strength of correlation is described using Pearson’s correlation coefficient. Significance was defined as p < 0.05. Data were analyzed with RStudio Version 2022.12.0 + 353 and Microsoft Excel Version 16.56.
In this study, we included 23 patients with a mean age of 51 (±13.1) years. Mean registered sequential organ failure assessment (SOFA) score was 10 (±4.7). Demographic and clinical baseline characteristics, as well as ECMO and ventilator settings were obtained from routine intensive care unit monitoring or from our patient data management system (Copra System GmbH, Berlin, Germany). Data are presented in Tables 1 and 2.
Of 150 obtained datasets, we excluded 17 datasets because of missing data or invalidity of IC or BGA values. We additionally considered a difference between pre- and postmembrane Hb greater than 0.5 g/dl and a respiratory quotient below 0.67 or greater than 1.3 to indicate invalidity. Furthermore, 11 datasets of patients undergoing ECMO weaning without GFML were excluded (see Supplement Figure 1, Supplemental Digital Content, http://links.lww.com/ASAIO/B227).
Calculations of CbloodO2 in arterial, pre-, and postmembrane blood show that dissolved oxygen accounts for a small amount in arterial (CaO2 dissolved: mean 2.5 ± 0.72 ml/l) and premembrane blood (Cpre-MLO2 dissolved: mean 1.04 ± 0.16 ml/l). Mean dissolved blood oxygen content in postmembrane blood (Cpost-MLO2 dissolved) was 11.85 ± 2.08 ml/l. This equates to 9.1% of the total postmembrane oxygen content (Figure 1).
Figure 1. Content of dissolved and hemoglobin bound oxygen in arterial (C
aO2 dissolved, CaO2 Hb, ml/l), pre- (Cpre-MLO2 dissolved, Cpre-MLO2 Hb, ml/l), and postmembrane blood (Cpost MLO2 dissolved, Cpost MLO2 Hb, ml/l).
In this study, group mean VO2 total was 269.16 ± 73.41 ml/min, of which 162.68 ± 49.49 ml/min were realized by extracorporeal oxygen transfer (VMLO2) and 106.48 ± 70.09 ml/min by the native lung (VNLO2). Mean VMLO2 dissolved was 36.41 ± 9.80 ml/min and VMLO2 Hb 126.27 ± 44.39 ml/min. This means that 22.4% of extracorporeal and 13.5% of overall oxygen uptake occurred as physically dissolved oxygen (Figure 2).
Figure 2. Shares of oxygen in extracorporeal and total oxygen transfer. A: Shares of dissolved (V
MLO2 dissolved) and hemoglobin bound (VMLO2 Hb) oxygen uptake in extracorporeal transfer (VMLO2). B: Shares of dissolved (VMLO2 dissolved) and hemoglobin bound (VMLO2 Hb) oxygen uptake in extracorporeal transfer and oxygen uptake by the native lung (VNLO2) in total oxygen transfer (VO2 total).
To evaluate the relationship between independent variables and oxygen uptake, we performed correlation analysis and found that VMLO2 was significantly determined by BFML and SO2 pre. According to expectations, GFML did not correlate with VMLO2. Furthermore, VMLO2 per liter ECMO blood flow (VMLO2/ BFML, ml/l) was found to be depended on Hb concentration. There was no statistically significant correlation between VMLO2 and PpostO2, which suggests that PpostO2 inadequately reflects the oxygen transport capacity of the ML (Supplement Figure 2, Supplemental Digital Content, http://links.lww.com/ASAIO/B227).
VMLO2 dissolved was strongly associated with BFML and showed moderate strong correlation with PpostO2, while we found only a weak association with GFML and none with SO2 pre (Figure 3). We analyzed the proportion that dissolved oxygen has in the total extracorporeal oxygen transfer and found that this fraction correlated well with PpostO2 and SO2 pre. Further increase in BFML and GFML did not affect VMLO2 dissolved pct (Figure 4).
Figure 3. Scatterplots showing the relationship between extracorporeal dissolved oxygen uptake in total extracorporeal oxygen uptake (V
MLO2 dissolved, ml/min) and the variables ECMO blood flow (BFML, l/min), ECMO sweep gas flow (GFML, l/min), premembrane oxygen saturation (SO2 pre, %), and postmembrane oxygen partial pressure (PpostO2, mmHg), with Pearson’s correlation r, R R^2^. A: VMLO2 dissolvedand BFML: r = 0.686, p < 0.001. B: VMLO2 dissolvedand GFML: r = 0.258, p = 0.009. C: VMLO2 dissolvedand SO2 pre: r = 0.124, p = 0.216. D: VMLO2 dissolvedand PpostO2: r = 0.334, p < 0.001. ECMO, extracorporeal membrane oxygenation.
Figure 4. Scatterplots showing the relationship between the proportion of dissolved oxygen uptake in total extracorporeal oxygen uptake (V
MLO2 dissolvedpct, %) and the variables ECMO blood flow (BFML, l/min), ECMO sweep gas flow (GFML, l/min), premembrane oxygen saturation (SO2 pre, %), and postmembrane oxygen partial pressure (PpostO2, mmHg), with Pearson’s correlation r, R R^2^. A: VMLO2 dissolvedpct and BFML: r = −0.039, p = 0.697. B: VMLO2 dissolvedpct and GFML: r = 0.000, p = 0.999. C: VMLO2 dissolvedpct and SO2 pre: r = 0.486, p < 0.001. D: VMLO2 dissolvedpct and PpostO2: r = 0.496, p < 0.001. ECMO, extracorporeal membrane oxygenation.
VMLO2 Hb primarily depended on BFML and lower SO2 pre was associated with an increase in VMLO2 Hb. GFML did not affect VMLO2 Hb. VMLO2 Hb per liter ECMO blood flow (VMLO2 Hb/BFML, ml/l) was moderately associated with Hb concentration (Supplement Figure 3, Supplemental Digital Content, http://links.lww.com/ASAIO/B227). Compared with the fraction of dissolved oxygen, the fraction of oxygen transferred bound to hemoglobin (VMLO2 Hb pct) only slightly increased with higher BFML. VMLO2 Hb pct did not depend on GFML and showed moderate negative correlation with SO2 pre (Supplement Figure 4, Supplemental Digital Content, http://links.lww.com/ASAIO/B227).
Both fractions of VMLO2 (VMLO2 dissolved and VMLO2 Hb) accounted for a higher percentage in VO2 total in patients with severely impaired oxygen uptake by the native lung (Figure 5).
Figure 5. Scatterplot showing the relationship between the proportion of extracorporeal dissolved and hemoglobin bound oxygen uptake in total oxygen uptake and oxygen uptake by the native lung (V
NLO2, ml/min); Pearson’s correlation r, R R^2^. r (dissolved) = −0.780, p < 0.001. r (Hb) = −0.881, p < 0.001.
In this study, we investigated to which extent dissolved oxygen (VMLO2 dissolved) partakes in extracorporeal oxygen transfer in adult ARDS patients treated with VV ECMO. The data demonstrate that in mean 22.4% of extracorporeal (VMLO2) and 13.5% of total oxygen uptake (VO2 total) were transferred as dissolved oxygen in our study cohort.
Because of its poor solubility (0.00314 ml/dl/mmHg), dissolved oxygen is often found to be negligible when assessing oxygenation during ECMO support.^6,20^ But when the inspiratory oxygen fraction (FiO2, %) of the GFML during VV ECMO support is set to 100%, PO2 and resultingly the amount of dissolved oxygen increases.
Spinelli and Bartlett^11^ published a mathematical model in 2014 to describe the determinants of extracorporeal oxygen uptake. They assume that simplifying the calculation of oxygen content by neglecting the share of dissolved oxygen leads to false results. Concomitantly, in a study, that was originally designed to validate a model for the assessment of recirculation in VV ECMO, Walker et al. come to a similar conclusion.^10^ They point out that due to high PO2, the share of dissolved oxygen in total oxygen content can be as much as 9%. Consistent with those studies, our calculations of postmembrane oxygen content show that the fraction of dissolved oxygen increases from 1.3% in premembrane blood to 9.1% in postmembrane blood, as a result of high PO2 (mean PpostO2 434.19 ± 76.29 mmHg).
While a mean of 36.41 ± 9.80 ml/min of VMLO2 dissolved
via ECMO seems to be only a tiny fraction, it needs to be considered that depending on the remaining function of the native lung, this can be a substantial share of overall oxygen delivery.
The patients’ arterial saturation results from the mixture of the oxygenated blood from the ECMO circuit with the venous blood, and depending on the residual lung function, further oxygen uptake by the native lungs. The surplus of dissolved oxygen from the ECMO presumably binds to deoxygenated Hb in the right atrium, thereby partly offsetting the effects of venous admixture.
In patients with severely impaired lung function, reflected by low VNLO2, extracorporeal oxygen uptake accounted for a higher proportion in VO2 total. Both fractions (VMLO2 dissolved and VMLO2 Hb) increased, while the ratio between them remained constant (Figure 5). A subgroup analysis of 40 datasets of patients with VNLO2/kg less than 1 ml/kg/min showed that up to 28% of VO2 total was covered as dissolved oxygen (mean 18.1%, 10–28.6%), which needs to be considered a significant share of overall oxygen supply.
Clinicians often rely on surrogate parameters such as PaO2 or SO2 to assess the patient’s oxygenation status and PpostO2 to detect ML failure.^21^ Our data demonstrate that PpostO2 does not correlate well with VMLO2 (Supplement Figure 2d, Supplemental Digital Content, http://links.lww.com/ASAIO/B227) and may not be suitable to detect pending ML failure due to clotting or fouling.
Recently developed monitoring devices advocate real-time monitoring of extracorporeal gas exchange may improve management of ECMO settings adjusted to the patient’s individual needs, guide through the process of ECMO weaning and surveille ML function. They are equipped with noninvasive clamp-on sensors for measurement of SO2, Hb, and BFML that can easily be connected to any ECMO circuit, but their technical properties do not allow measurement of PO2.^22,23^ Consequently, the content of dissolved oxygen in pre- and postmembrane blood cannot be calculated, leading to an underestimation of VMLO2. Accurate knowledge of the oxygenation capacity is crucial for making therapy decisions related to managing ECMO settings or replacing the ML. The Terumo CDI System 500 (Terumo Cardiovascular, Ann Arbor, MI), a blood gas analyzing system, that was developed for monitoring of cardiopulmonary bypass, includes inline shunt sensors, which are capable of measuring PO2, among other blood gases. However, those sensors are not approved for long-term continuous usage.^24^ The CARL system (Resuscitec GmbH, Freiburg, Germany), an extracorporeal cardiopulmonary resuscitation (eCPR) system recently developed for targeted reperfusion in out-of-hospital cardiac arrest, provides a sensor technique that allows continuous assessment of blood gases. Similarly, to the sensors integrated in the Terumo CDI System 500 time of use is limited.
While it is technically possible to measure blood gases in extracorporeal circuits, limitations regarding the duration of use, make inline sensors unfit for usage in long-term VV ECMO support.
Extracorporeal oxygen transfer (VMLO2) is determined by multiple factors rather than one. While most properties of the ML itself cannot be influenced during an ongoing VV ECMO therapy, clinicians need to be aware of factors they can alter, if they want to adjust VMLO2.
In line with other studies, the data from our patient cohort demonstrate that under clinical conditions VMLO2 was significantly determined by BFML, SO2 pre, and Hb.^17,25^ Comparing the influence of SO2 pre on VMLO2 dissolved pct and VMLO2 Hb pct, this study shows that when Hb is fully saturated, a further increase in VMLO2 can only be realized as dissolved oxygen (Figure 4C, Supplement Figure 4c, Supplemental Digital Content, http://links.lww.com/ASAIO/B227). The uptake of dissolved oxygen might especially be of relevance for those patients with lower Hb to compensate for the reduced oxygen transport capacity.
As expected, GFML did not affect VMLO2 if an FiO2 of 1.0 is used and even during ECMO weaning, when GFML is reduced to 0.5 L/min (while BFML remains ≥2.0 L/min and FiO2 is kept at 100%), up to 230 ml/min of oxygen can still be transferred via the ML. This implicates that the sole reduction of the sweep gas flow rather weans the patient from extracorporeal CO2 elimination than from extracorporeal oxygenation. Thus, a reduction of the oxygen fraction within the sweep gas, followed by a complete, prolonged cessation of the sweep gas flow is necessary to evaluate patients’ pulmonary function before ECMO decannulation.^26^ Continuous monitoring of VMLO2 supplemented by surveillance of the function of the native lung can help identifying the optimal point of time to proceed with ECMO weaning.
This research was conducted as a single centered observational study. We performed a noncontinuous measurement of VMLO2 in each individual patient and did not aim to depict changes in oxygen uptake over time. While assessing blood gases using by BGA is an accurate method to determine PO2, SO2, and Hb and subsequent calculation of VMLO2, our dropouts indicate that the preanalytical handling of blood samples is prone to cause errors, even when used by experienced staff. For the measurement of VNLO2, we used IC, which has, due to its technical setup, limitations regarding its applicability in critically ill, mechanically ventilated patients. Beside the requirement of a steady state, measurements are limited to, for example, a FiO2 below 75%, and therefore does not allow assessment of the native lung function in patients with severe ARDS, excluding this specific cohort.
In patients with minimal residual lung function where oxygenation almost or completely depends on extracorporeal oxygen transfer, exact knowledge of oxygen uptake via ECMO and by the native lung is essential for managing ECMO settings.
As dissolved oxygen contributes in a clinically relevant amount to the oxygenation in patients undergoing VV ECMO support, it must be considered when evaluating the extracorporeal oxygen supply and ML function. Therefore, devices for continuous monitoring of ECMO therapy should incorporate a sensor technique that allows measurement of the whole blood oxygen content.