Effect of Hematocrit on the CO₂ Removal Rate of Artificial Lungs

Abstract

Extracorporeal CO₂ removal (ECCO₂R) can permit lung protective or noninvasive ventilation strategies in patients with chronic obstructive pulmonary disease (COPD) and acute respiratory distress syndrome (ARDS). With evidence supporting ECCO₂R growing, investigating factors which affect CO₂ removal is necessary. Multiple factors are known to affect the CO₂ removal rate (vCO₂) which can complicate the interpretation of changes in vCO₂; however, the effect of hematocrit on the vCO₂ of artificial lungs has not been investigated. This in vitro study evaluates the relationship between hematocrit level and vCO₂ within an ECCO₂R device. In vitro gas transfer was measured in bovine blood in accordance with the ISO 7199 standard. Plasma and saline were used to hemodilute the blood to hematocrits between 33% and 8%. The vCO₂ significantly decreased as the blood was hemodiluted with saline and plasma by 42% and 32%, respectively, between a hematocrit of 33% and 8%. The hemodilution method did not significantly affect the vCO₂. In conclusion, the hematocrit level significantly affects vCO₂ and should be taken into account when interpreting changes in the vCO₂ of an ECCO₂R device.

Patients with moderate acute respiratory distress syndrome and acute exacerbations of chronic obstructive pulmonary disease can benefit from lower tidal volume, lung protective ventilation, or noninvasive ventilation strategies.^1–5 These ventilation strategies can prevent ventilator-induced lung injuiy; however, failure is commonly due to hypercapnia and the resulting acidosis and dyspnea.⁶ Low-flow extracorporeal CO₂ removal (ECCO₂R) devices have been used in this patient population to remove the excess CO₂ while maintaining lung protective or noninvasive ventilation.^1,7 ECCO₂R devices operate on analogous principles to extracorporeal membrane oxygenation. Because of the primary objective of removing CO₂ rather than oxygenation, ECCO₂R devices can operate at blood flow rates below 1 L/min.⁸ The potential benefits of ECCO₂R are currently being evaluated in the VENT-AVOID (clinicaltrials.gov: NCT03255057) and REST (clinicaltrials.gov: NCT02654327) clinical trials. ECCO₂R device selection and operating parameters dictate the CO₂ removal rate (vCO₂) but the nuances of CO₂ removal can make interpreting changes in vCO₂ challenging. Correct interpretation is especially important as vCO₂ is the most sensitive indicator of intrabundle thrombosis.⁹

Numerous factors affect the CO₂ transfer rate within native and artificial lungs.¹⁰ The vCO₂ of a specific artificial lung geometry is a known function of blood flow rate, inlet pCO₂, and sweep gas flow rate. The effect of extracorporeal blood flow rate and inlet pCO₂ on vCO₂ can be taken into account by normalizing the vCO₂ to these parameters.¹¹ The effect of sweep gas flow rate on vCO₂ and the importance of a sweep gas independent regime have also been well established.¹² Diminished vCO₂ performance, however, has also been noted in animal studies during periods of anemia.^13–15 The relationship between vCO₂ and hematocrit (HCT) seen in artificial lungs has been attributed to mechanisms analogous to those within the native lungs. Bidani and Crandall¹⁵ used computational modeling to demonstrate that a decrease in HCT from 45% to 15% results in a 50% decrease in vCO₂ within the native lungs. A similar vCO₂–HCT relationship has been noted in animal studies;¹³ however, the complexities of animal studies make isolating the artificial lung vCO₂–HCT relationship challenging.

In this study, we evaluated the relationship between HCT and the vCO₂ of a low-flow ECCO₂R device in an in vitro setting. Blood was hemodiluted with saline and plasma to reach the targeted HCT to evaluate any effect plasma proteins may have on vCO₂. The in vitro setting allowed the effect of HCT and plasma proteins on vCO₂ to be isolated. The results of this study will aid in predicting vCO₂ in ECCO₂R devices and interpreting changes in vCO₂ related to HCT.

Methods

In vitro Gas Exchange

Bovine blood was collected from a local slaughterhouse and was heparinized (30 IU/ml), filtered (40 μm; Pall Biomedical, Inc., Fajardo, PR), and treated with gentamicin (0.1 mg/ml). Blood was diluted to HCT levels of 8, 15, 25, and 33 ± 2% with either isotonic saline (0.9% [w/v] Aqueous, Isotonic Saline; Fisher Scientific, Hampton, NH) or plasma. Plasma was obtained by centrifuging the whole bovine blood (2,819g for 15 minutes) and removing the plasma layer.

Gas exchange testing followed the ISO7199 standard¹⁶ with the exception of the intentional alterations in HCT levels. The test circuit (Figure 1) included the Modular Extracorporeal Lung Assist System (ModELAS) artificial pump-lung (0.65 m² hollow fiber membrane bundle),¹¹ two 6 L blood reservoir bags, and an Affinity oxygenator (Medtronic, Minneapolis, MN). The reservoir bags were submerged in a water bath to maintain a blood temperature of 37 ± 1°C. The Affinity oxygenator was used to condition the blood to venous blood gas tensions (sO₂ = 65 ± 5%, pCO₂ = 45 ± 5 mmHg). Once the blood reached venous conditions, tubing clamps were manipulated so that blood was drawn from the venous conditioned reservoir, through the device, and returned to the second reservoir. Samples for blood gas measurements were taken immediately before and after the ModELAS. A RapidLab 405 blood gas analyzer was used to measure gas tensions (Siemens, Deerfield, IL). The CO₂ removal rate, hemoglobin concentration, and plasma protein concentration were measured at each HCT level in triplicate.

Figure 1.

Schematic depicting the blood and gas flow pathways of the in vitro gas exchange loop.

The pump speed was adjusted to achieve 500 ml/min of blood flow, typical for low-flow ECCO₂R devices,^17,18 through the ModELAS. Blood flow rate was measured using an ultrasonic flow probe (Transonic Systems Inc., Ithaca, NY). The pure oxygen sweep gas flow rate (Q_SG) was set to 14 L/min using a gas flow controller (Fathom Technologies, Georgetown, TX). From preliminary experiments, this sweep gas flow rate is within the sweep gas independent regime. CO₂ concentration (F_CO2) was measured in the exhaust sweep gas using a WMA-4 gas analyzer (PP Systems, Amesbury, MA). The CO₂ removal rate was calculated and normalized (${\text{vCO}}_2^{\ast }$) to an inlet pCO₂ of 45 mm Hg according to Equation 1.

$${\text{vCO}}_{\text{2}}^{\text{∗}}=Q_{SG}{F}_{C{O}_2}\frac{45\text{mmHg}}{P_{C{O}_2}^{Inlet}}$$ (1)

An additional 4 ml blood sample was drawn from the device inlet to measure HCT and plasma protein each time a CO₂ removal measurement was taken. HCT was measured using a capillary tube which was spun down for 3 minutes in a microcentrifuge (International Equipment Co., IEC MB, Needham Hts, MA). The remaining blood was centrifuged for 15 minutes at 0.8g, and the supernatant plasma was removed and centrifuged again at 7.2g for 10 minutes. The plasma protein concentration was measured using a refractometer (TS 400, Reichert, Depew, NY) from this purified plasma.

Statistical Analysis

Data are reported as the mean ± standard deviation. Statistical analysis was completed in SPSS (IBM, Armonk, NY). A Friedman’s analysis of variance was conducted to evaluate the effect of dilution method and HCT on measured vCO₂, plasma protein concentration, pH, and calculated bicarbonate concentration. Pairwise comparisons were made using Wilcoxon tests. Statistical significance was determined at p < 0.05.

Results

In vitro vCO₂ for plasma and saline diluted blood are shown in Figure 2. There was no significant effect of dilution method (plasma dilution or saline dilution) on vCO₂ (p = 0.564). The CO₂ significantly decreased (p = 0.000) with decreasing HCT in a nearly linear fashion for both saline dilution (R² = 0.998) and plasma dilution (R² = 0.984). Decreasing the HCT from 33% to 8% resulted in a 32% and 42% decrease in vCO₂ when diluted with plasma and saline, respectively. The plasma protein concentration (Figure 3) decreased significantly after each hemodilution with saline (p = 0.001) and was significantly different between plasma and saline dilution (p < 0.05). Inlet and outlet bicarbonate ion concentrations were not affected by the dilution method (Figure 4; p = 0.248 and p = 0.083, respectively). The inlet pH was not affect by the dilution method (p = 0.083); however, the dilution method substantially affected the device outlet pH (p = 0.001) (Figure 4).

Figure 2.

CO₂ removal rates measured at hematocrit increments of 8, 15, 25, and 32 ± 2% for blood diluted with saline or plasma. The blood flow rate through the device was held constant at 0.5 L/min.

Figure 3.

Plasma protein concentration of the circulating blood after hemodilution with saline or plasma. Hemodilution with saline significantly decreased the plasma protein concentration at all hematocrit levels p < 0.05).

Figure 4.

Calculated ${\text{HCO}}_3^{-}$ concentration and measured pH of device inlet and out blood at each hemodilution level. ${\text{HCO}}_3^{-}$ concentration was calculated using the Henderson-Hasselbalch equation¹⁹.

Discussion

The evidence supporting the use of ECCO₂R to permit noninvasive or lung protective ventilation strategies in patients with moderate acute respiratory distress syndrome and acute exacerbations of chronic obstructive pulmonary disease continues to increase. With this comes a need to identify variables which influence the vCO₂ of artificial lungs, especially because a decrease in vCO₂ can be indicative of intrabundle thrombosis.⁹ The vCO₂ of an artificial lung can be affected by blood and sweep gas flow rates, and the inlet pCO₂ of the blood.^11,12 The effect of HCT on CO₂ transfer has been documented in computational studies within the native lungs but only observationally noted in artificial lungs during in vivo studies.¹³ This in vitro study evaluated the vCO₂ within an ECCO₂R device in blood diluted to multiple HCT levels using plasma and saline. In vitro testing demonstrated that the vCO₂ linearly and significantly decreased with decreasing HCT level. CO₂ removal rate was not affected by the method of hemodilution.

CO₂ transfer decreases with the decrease in HCT within artificial lungs, as demonstrated here, as well as within the native lungs.^14,15,20,21 The vCO₂–HCT relationship within the native lungs has been attributed to 1) decreased buffering capacity provided by hemoglobin, 2) reduced flux of ${\text{HCO}}_3^{-}$ across the red blood cell (RBC) membrane, and 3) the release of fewer Bohr protons.¹⁵ The first two mechanisms are a direct result of fewer RBCs. Carbonic anhydrase (CA) within the RBC plays an integral role in rapidly hydrating CO₂ via the catalysis of ${\text{CO}}_2{+\text{H}}_2\text{O}{\iff \text{HCO}}_3^{-}{+\text{H}}^{+}$. Hemoglobin present within the RBC acts as a buffer by binding to H⁺. In addition, the ${\text{Cl}}^{-}{\text{/HCO}}_3^{-}$ RBC membrane-bound exchanger rapidly exchanges ${\text{HCO}}_3^{-}$ for Cl⁻, the chloride shift, to prevent the rate-limiting accumulation of ${\text{HCO}}_3^{-}$ within the RBC. These two mechanisms prevent the accumulation of H⁺ and ${\text{HCO}}_3^{-}$ within the RBC which shifts equilibrium to the right. Reductions in hemoglobin and RBC concentration diminish the buffering capacity and the surface area available for the chloride shift to occur. Thus, the reaction reaches equilibrium sooner and less CO₂ is combined with blood per increase in pCO₂. The third mechanism is the classic Haldane effect whereby the uptake of oxygen by hemoglobin in the lungs decreases the affinity of hemoglobin for CO₂ and results in the release of H⁺ and bound CO₂. The released H⁺ drives the equilibrium of the reaction to the left, toward dissolved CO₂. Although less than 5% of the total CO₂ content within blood is transported bound to hemoglobin, this source accounts for 30% of exhaled CO₂.²² Bidani and Crandall¹⁵ used a computational model to isolate each of these mechanisms and demonstrated that 50–60% of the decrease in vCO₂ because of decreased HCT is attributable to the third mechanism. Thus, even moderate decreases in HCT will result in decreased CO₂ removal. These three factors result in anemic blood dissociating with less CO₂ per decrease in pCO₂. This essentially results in a flattening of the CO₂ dissociation curve as the HCT level decreases.¹⁴ The above three mechanisms are inherent to properties of the blood, and not of the native lungs. Thus, these mechanisms are likely identical to those influencing the vCO₂–HCT relationship within the artificial lung.

The effects of anemia on CO₂ exchange within the native lungs is often masked by compensatory mechanisms within the body. In this study, however, a 25% decrease in HCT led to a 13% decrease in vCO₂. Normal gas exchange within the body is primarily maintained during periods of anemia by increases in cardiac output and ventilation rate, and changes to the microcirculation and metabolism.²⁰ Within an artificial lung, decreases in HCT will result in increased extracorporeal blood flow rate if the pump speed is held constant and would be analogous to increased cardiac output. In the current study, however, the pump speed was adjusted to maintain an extracorporeal blood flow rate of 500 ml/min as the HCT level was decreased. The other three compensatory mechanisms would not be mimicked within an artificial lung. Hence, even small changes to the HCT level affected the vCO₂ of the artificial lung during this study.

The effect of plasma proteins was evaluated in the current study by comparing the vCO₂ of blood diluted with plasma versus that diluted with normal saline. Plasma proteins, mainly albumin and phosphates, typically contribute minimally to CO₂ transport and CO₂ removal. This is becasue of the ability of normal levels of hemoglobin to buffer approximately six times more H⁺ than plasma proteins are capable of.²³ In anemic blood, however, the role of the buffering action of plasma proteins has been postulated to be more significant because of the reduced amount of hemoglobin.¹⁴ There was not a significant effect of dilution method (saline versus plasma) on vCO₂ in this study, even though there was a significant difference in plasma protein concentration at each HCT level between the two dilution methods. Arazawa et al.²⁴ report a similar in vitro result, whereby a miniature artificial lung only removed 13 ml/min/m² more CO₂ in a phosphate buffered saline+albumin solution than in a phosphate buffered saline solution. The results of the current study, and that of Arazawa et al., demonstrate that the buffering capacity of plasma proteins is not significant enough to accommodate for the reduction, or absence, of HCT within an artificial lung. This is because of the CO₂ carrying capacity of plasma being substantially due to the chloride shift, which requires the presence of the RBC, to occur.²³ Thus, regardless of the nature of anemia (hemodilution versus diminished RBC production), an equivalent decrease in artificial lung vCO₂ should be anticipated as HCT level decreases.

The device was operated at a high sweep gas flow rate to achieve sweep gas independence and thereby isolate the effect of HCT on vCO₂. Sweep gas flow rate and CO₂ removal are more inter-related than oxygenation and sweep gas flow rate. At low sweep gas flow rates, CO₂ begins to accumulate within the sweep gas as it passes through the lumen of the fiber. This causes the CO₂ concentration driving gradient between the blood and sweep gas to become diminished and reduces the overall CO₂ removal of the device. Thus, to remove the influence of CO₂ build-up within the fiber lumen, benchtop ECCO₂R experiments are typically operated in a sweep gas independent regime. Clinically, the sweep gas flow rate may be adjusted to achieve the level of CO₂ removal the patient requires and even steadily decreased to wean patients off of the device. By implication, a clinician would be lowering sweep gas because the extracorporeal CO₂ removal is more than adequate and hence the effect of HCT on extracorporeal CO₂ removal would be less relevant. In the present benchtop study, the goal was to isolate the effect of HCT; thus, the device was operated within a sweep gas independent regime.

In this study, as the blood was diluted, the viscosity of the blood consequently also decreased. Extracorporeal blood flow rate was maintained at 500 ml/min by reducing the pump speed after each dilution. In an artificial lung, fluid viscosity (ν) affects the fluid boundary layer thickness at the fiber surface by ν^1/6. The fluid side boundary layer provides the majority of the diffusional resistance to gas transfer within the artificial lung. Thus, with all else held constant, the gas transfer rate will increase as fluid viscosity decreases within an artificial lung.¹⁰ The results described within this article do not make a distinction between increases in vCO₂ due to a lower viscosity fluid and diminished vCO₂ due to lowered HCT level.

The interpretation of changes in vCO₂ is only reliable if all factors influencing vCO₂ are accurately accounted for. The results of this in vitro study provide evidence demonstrating that decreases in HCT result in decreased artificial lung CO₂ transfer regardless of the nature of the anemia. Changes in the HCT must therefore be taken into account when determining the cause of changes in vCO₂ within an artificial lung.