The term extracorporeal carbon dioxide removal (ECCO2R) describes the use of gas exchange devices to directly eliminate CO2 from the blood stream.1 Extracorporeal carbon dioxide removal is used to facilitate lung-protective ventilation in patients with asthma or acute respiratory distress syndrome,2,3 and may even provide an alternative to intubation in patients with chronic obstructive pulmonary disease (COPD).4,5 Depending on the surface area available for gas exchange, commercially available ECCO2R devices typically remove 90–150 ml·min−1 of CO2, using blood flow rates ranging from 400 to 1500 ml·min−1.6–8
Extracorporeal carbon dioxide removal was first used over 30 years ago,9,10 but its use remains mostly restricted to specialist centers that can place large cannulas and manage complications.1 Although less expertise is required for newer low-flow ECCO2R devices, which use blood flows less than 500 ml·min−1, they are expensive and not widely available. In contrast, dialysis devices are readily available in most intensive care units,11,12 and since CO2 is primarily transported as bicarbonate,13 bicarbonate dialysis may represent an alternative strategy for ECCO2R. Previous attempts to provide bicarbonate dialysis were limited by electrolyte and acid-base derangements, hemolysis, cardiac arrhythmias, or complications because of “replacing” bicarbonate with various metabolizable anions.14–17 Fortunately, it may not be necessary to replace bicarbonate at all while avoiding acid-base derangements, provided the relative concentrations of plasma strong ions are maintained.
Strong ions are defined as molecules which completely dissociate when dissolved in water, such as sodium and chloride. Contemporary understanding of plasma acid-base predicts that plasma bicarbonate concentration is dependent on the amount of dissolved CO2, the difference between concentration of dissolved strong cations and anions (strong ion difference or SID), and total weak acid (Atot, mostly composed of protein).18 This concept is frequently described as the physical chemistry approach, or Stewart approach, in recognition of Peter A. Stewart19 who published the concept in 1981. Recently, the underlying model in the physical chemistry approach was developed as an explicit notion of charge balance, and successfully applied to both experimental data and intensive care patients.20
We hypothesized that a dialysate solution with a low bicarbonate concentration will remove CO2 from plasma and increase pH, without attempts to replace the bicarbonate, provided the electrolyte composition of the dialysate ensures SID and Atot are maintained within the physiologic range. We developed a mathematical model and conducted in-vitro dialysis experiments to test this hypothesis.
Materials and Methods
We used the physical chemistry understanding of acid-base to build a model capable of predicting the effect of bicarbonate dialysis on plasma CO2 levels and pH. A complete derivation of the theoretical model is provided in the Methods section in Supplemental Digital Content (http://links.lww.com/ASAIO/A339), but the summary equation, based on charge balance, is provided here to facilitate understanding of the results.
[SID] (“strong ion difference”) is the difference between concentrations of totally dissociated cations and anions (mEq·L−1), [H+] is the concentration of hydrogen ions (mEq·L−1), Kc is the combined equilibrium and solubility constant for CO2, Pc is the partial pressure of CO2 (mm Hg), [Atot] is the total weak acid (mEq·L−1), Ka is the combined dissociation constant for Atot, and K′w is the dissociation constant for water.
To develop the model, we made several assumptions. First, we assumed CO2 is not ventilated to, or absorbed from, the atmosphere at the level of the dialysis filter. Second, we assumed that the dialysate would have the same SID as normal plasma. Finally, we assumed no protein was dialysed out of solution or adsorbed to the filter. Therefore, SID and Atot could be held at constant standard values in the model (Table 1),21,22 allowing us to solve Equation 1 for a range of partial pressures of CO2 (PCO2) values using an iterative procedure to determine [H+] (and hence pH). The equation was solved for 10,000 individual CO2 values, ranging from 0.1 to 100 mm Hg. Bicarbonate was subsequently determined using PCO2 and [H+] (see Methods, Equation ES2, Supplemental Digital Content, http://links.lww.com/ASAIO/A339).
In-Vitro Dialysis Experiment
We designed an in-vitro circuit where we could add CO2 to artificial plasma and then pump the plasma through a dialysis filter, allowing us to test the CO2 removal properties of three dialysates with different bicarbonate concentrations (Figure 1). The three dialysis solutions used were Prismasol (Gambro, Lyon, France), Bicarb16, and Bicarb0. Prismasol served as our conventional (control) dialysate; its bicarbonate (HCO3) concentration is 32 mmol·L−1 (Table 2). Bicarb16 and Bicarb0 were constructed to have lower HCO3 concentrations (16 and 0 mmol·L−1, respectively) by adding salts to deionized water. All three dialysates had similar target electrolyte compositions, but we adjusted final pH down to 8.0 by adding hydrochloric acid. Therefore, final chloride concentrations varied (Table 2). We created an artificial plasma solution by adding 0.544 g (4 mmol) of monopotassium phosphate (KH2PO4) to 1000 ml of Prismasol (Gambro, Lyon, France), the resulting electrolyte concentrations reflected normal plasma (Table 2). We added CO2 to our artificial plasma using a hollow fiber oxygenator (Minimax Plus, Medtronic, Anaheim, CA). The oxygenator sweep gas was composed of CO2 and nitrogen (N2), and the flow of each gas was adjusted with a gas blender to achieve target PCO2. The artificial plasma was then pumped through a M10 dialysis filter (Gambro, Lyon, Paris) at 10 ml·min−1 (Peristaltic pump, ISM833C, Reglo Digital, Ismatec, Wertheim, Germany), where exposure to dialysate occurred. Since the M10 dialysis filter has a surface area of 0.042 m2, our flow rate of 10 ml·min−1 is the equivalent of 214 ml·min−1 with an adult filter of surface area 0.9 m2 (filter scaling factor = 0.9/0.042 = 21.4, i.e., 10 × 21.4 = 214 ml·min−1).
The consequences of plasma exposure to the various dialysates were recorded by measuring electrolyte and gas composition of the plasma pre- and postfilter (Figure 1). Dialysate flow rates started at 0 ml·min−1 and were increased to 20 ml·min−1 by increments of 2 ml·min−1. A dialysate flow of 2 ml·min−1 provided the equivalent dialysis dose of 34 ml·kg−1·hr−1, using 75 kg as a reference body weight and the filter surface area scaling factor of 21.4 (2 × 21.4 × 60 ÷ 75).
Plasma samples were analyzed for pH, PCO2, and electrolyte concentrations. We repeated sample analysis every 5–10 minutes until three consecutive readings were similar at each dialysis flow rate, indicating system stabilization. Partial pressure of CO2 and pH were measured with an iStat CG3 cartridge (Abbott, Princeton, NJ). Only plasma PCO2 is reported because the iStat does not directly measure bicarbonate but calculates it from measured pH and PCO2. Electrolytes were analyzed by first diluting the samples to a concentration range of 5 ppm. The concentrations of metal ions (potassium, magnesium, calcium, and sodium) were measured by inductive coupled plasma-optical emission spectroscopy (ICP-OES), and the concentration of chloride ions were determined by anion-ion exchange chromatography (IC). Electrolyte analysis was conducted by the Element Analysis Laboratory (EAL) at the National University of Singapore.
Calculation of Carbon Dioxide Removal and Statistical Analysis
Estimated total CO2 removal, in milliliters per minute, for an adult-sized filter was converted from the measured change in PCO2 in the plasma entering and exiting the filter, using the following equation:
Δ[HCO3−] is the change in HCO3 concentration across the dialysis filter (mmol·L−1), ΔPCO2 is the change in PCO2 (mm Hg) across the filter, Ks is the CO2 solubility constant, 0.03 (mmol·mm Hg−1·L−1), Qf is the flow rate through experimental filter (L·min−1), Vm is the molar volume at standard temperature and pressure (22.4 ml·mmol−1).
HCO3 was calculated from pH and PCO2 using the following equation:
where [HCO3] is bicarbonate concentration in mmol·L−1 and PCO2 is partial pressure of CO2 in mm Hg. The data are presented as mean ± standard deviation, unless otherwise stated. Differences were compared using one-way analysis of variance (ANOVA) or ANOVA for repeated measures. Post-hoc multiple comparisons were performed using the Bonferroni test. A p value of <0.05 was considered statistically significant. All analysis was performed using STATA version 13 (College Station, TX).
Mathematical Bicarbonate Dialysis Modeling Experiment
Figure 2 shows the modeling experiment results where PCO2 is varied from 20 to 100 mm Hg, while concentrations of plasma SID and Atot are kept constant. Predicted pH decreases substantially (7.69–7.07) as PCO2 increases, while predicted plasma bicarbonate concentration only changes modestly (24–28.5 mmol·L−1). Figure 3 shows the results of a second modeling experiment, where bicarbonate is reduced in a system with no access to atmospheric CO2, while SID and Atot concentration are constant. This represents the effect of removing bicarbonate from blood as it passes through a dialysis filter, where exposure to a low bicarbonate dialysate occurs. The model predicts pH increases (7.07–11.37) and PCO2 decreases (100 to <0.1 mm Hg), when bicarbonate concentration is reduced from 28.5 to <1 mmol·L−1.
In-Vitro Bicarbonate Dialysis Experiment
Figure 4 shows measured PCO2 in our artificial plasma solution as it exits the dialysis filter (post filter), following exposure to our three different dialysis solutions: Prismasol, Bicarb16, and Bicarb0 (Table 2). The experiment was conducted using three different prefilter CO2 conditions, PCO2 50 mm Hg (50.64 mm Hg ± 2.93, pH 7.3 ± 0.64), PCO2 70 mm Hg (71.96 mm Hg ± 3.44, pH 7.16 ± 0.63), and PCO2 90 mm Hg (91.91 mm Hg ± 3.18, pH 7.12 ± 0.02). Postfilter PCO2 decreased with increasing dialysate flow rate for all dialysates (Figure 4). Under most experimental conditions, the change in PCO2 was not significantly different between Bicarb16 and Bicarb0, but both were significantly different from Prismasol (see Table S1, Supplemental Digital Content, http://links.lww.com/ASAIO/A340).
Table 3 shows estimated CO2 removal using a dialysis circuit with an adult dialysis filter (surface area 0.9 m2) and a blood flow rate of 214 ml·min−1. At all dialysate flow rates, the change in PCO2 was significantly higher for Bicarb0, compared with both Primasol and Bicarb16. Total CO2 removal with Bicarb16 was generally 45–65% of the removal seen with Bicarb0, except when target PCO2 was 50 mm Hg (Table 3). Total CO2 removal starts to plateau at a dialysate flow of 171 ml·min−1 (effluent dose, 137 ml·kg−1·hr−1 if 75 kg adult) for Bicarb16 and Bicarb0, as shown by no significant difference in total CO2 removal when dialysate flow is increased to 214 and 429 min·min−1 (Table 3), with the exception of Bicarb0 when prefilter PCO2 is 90 mm Hg. Maximum estimated CO2 removal, when dialysate flow equals plasma flow rate, is 53.4 (± 3.3) mL·min−1 for Bicarb16 and 93.6 (± 3.2) mL·min−1 for Bicarb0. This is seen when the driving gradient is highest, i.e., when prefilter target PCO2 is 90 mm Hg.
Figure 5 shows the postfilter pH for each dialysate under the three different target PCO2 conditions. Postfilter pH is similar for all three dialysates when no dialysate is flowing, except when target PCO2 is 50 mm Hg, where a slight difference in pH for Bicarb0 and Bicarb16 meets statistical significance (7.34 ± 0.1 vs. 7.37 ± 0.1; p = 0.04). Postfilter pH increases for all three dialysis solutions as dialysis flow rate is increased. However, postfilter pH rises slower for Bicarb0 when target PCO2 is 70 and 90 mm Hg, and remains significantly different from Prismasol at all flow rates (see Table S2, Supplemental Digital Content, http://links.lww.com/ASAIO/A341). In contrast, postfilter pH for Bicarb16 is statistically similar to that of Prismasol, particularly at lower dialysis flow rates (see Table S2, Supplemental Digital Content, http://links.lww.com/ASAIO/A341). Figure 6 shows the calculated postfilter SID at different flow rates for the three dialysis solutions. The SID remains constant over the full range of dialysis flow rate for Prismasol and decreases for Bicarb16 and Bicarb0. Measured electrolytes are shown in Table S3 (Supplemental Digital Content, http://links.lww.com/ASAIO/A342) and Figure S1 (Supplemental Digital Content, http://links.lww.com/ASAIO/A343), showing that postfilter chloride concentration follows the same trend as the SID for all three dialysates.
Mathematical modeling confirms our prediction of a decrease in PCO2 and an increase in pH when bicarbonate is removed by dialysis, despite no attempt to “replace” bicarbonate, provided the SID and Atot remain unchanged (Figure 3). When we translate this result into an in-vitro experiment, our data show that a low bicarbonate dialysate (Bicarb16) lowers PCO2 in a dose-dependent fashion (Figure 4) while increasing, or restoring, pH (Figure 5). Our data also show that pH is inadequately restored following bicarbonate removal with a dialysate which does not adequately control the SID (Bicarb0 buffered with hydrochloric acid; Figure 5), despite higher total CO2 removal (Table 3).
Mathematical modeling was conducted in two stages. In the first stage, we showed that plasma bicarbonate changed only modestly across a wide range of physiologic PCO2 values, when the SID and Atot remain constant (Figure 2). Similar behavior was predicted by Stewart,23 who showed that bicarbonate concentration changes little over a range of PCO2 values for interstitial fluid, provided the SID and Atot are unchanged. However, we also demonstrated that when bicarbonate is reduced, for example with bicarbonate dialysis, the model predicts a rapid reduction in PCO2, and a rise in pH, provided SID and Atot are held constant (Figure 3). This reflects the fact that dissolved CO2 is mostly carried in the form of bicarbonate.13 Therefore, removing bicarbonate results in substantial removal of CO2 from the system, ultimately increasing pH.
Previous CO2 removal experiments using bicarbonate dialysis assumed bicarbonate to be the principle determinate of pH, and therefore attempted to replace it. In 1991, Nolte et al.15 searched for a suitable replacement anion to support bicarbonate dialysis. They tested sodium hydroxide, tris(hydroxymethyl)aminomethane (THAM), and organic anions (lactate, acetate, pyruvate, malate, succinate, fumarate, and citrate). The authors reported that sodium hydroxide infusions resulted in electrolyte disturbances, hemolysis, cardiac arrhythmias, and cardiac arrest, whereas organic anions did not support a stable pH or were metabolized to CO2. THAM worked without side effects for up to 8 hours, after which intractable hypotension developed.15 More recently, hemofiltration has been used to remove bicarbonate. These experiments used sodium hydroxide in the replacement fluid, which also had a high chloride content. As a result, hyperchloremic acidosis developed.16 Our modeling data show it is not necessary to replace to bicarbonate, provided the dialysate is engineered to maintain the SID within acceptable limits, potentially eliminating the complications observed with “bicarbonate replacement” in earlier studies.
Using an in-vitro design, we translated our mathematical model into a simple bench experiment. We showed that low bicarbonate dialysis solutions are capable of lowering PCO2 (Figure 4). It was particularly interesting to observe that our control dialysis solution (Prismasol) also lowered PCO2, especially when prefilter target PCO2 was 90 mm Hg. Similar effects of conventional dialysis on PCO2 have been observed before.24 In fact, the development of membrane lungs was inspired by the observation that gas exchange occurred across cellophane tubing used in early dialysis devices.25 When acetate is used as the dialysate buffer, instead of bicarbonate, PCO2 can be reduced by over 10 mm Hg in patients with a resting PCO2 of 35 mm Hg.26 However, acetate dialysis is no longer favorable because of complications such as hypotension and nausea.27–29Table 3 shows that the PCO2 changes observed with Prismasol did not translate into actual CO2 removal, this is because bicarbonate concentration was essentially unchanged when dialyzing with Prismasol.
Our low bicarbonate dialysates (Bicarb16 and Bicarb0) resulted in much greater postfilter PCO2 reductions. As expected, the reduction in postfilter PCO2 increased as dialysate flow rate increased, reaching a plateau when dialysate flow rate approached the plasma flow rate (10 ml·min−1, see Figure 4). It is noteworthy that the change in PCO2 is similar for both low bicarbonate dialysis solutions (see Table S1, Supplemental Digital Content, http://links.lww.com/ASAIO/A340). Bicarb0 contains no bicarbonate, whereas Bicarb16 contains a bicarbonate concentration of 16 mmol·L−1, therefore more CO2 removal is expected with Bicarb0. This is indeed observed when total CO2 removal is considered (Table 3). However, our observation can be explained by considering the modeling results in Figure 3, where PCO2 changes very little once the bicarbonate concentration is less than 20 mmol·L−1. As the bicarbonate concentration in both Bicarb16 and Bicarb0 is < 20 mmol·L−1, our modeling predicts postfilter PCO2 should be similar for both dialysates, exactly what Figure 4 demonstrates.
However, Figure 3 would also predict a higher postfilter pH with Bicarb0, because postfilter bicarbonate should be lower (and hence total CO2 also lower) than that for Bicarb16. In our experiment, this is clearly not the case. In fact, Figure 5 shows that postfilter pH is lower for Bicarb0. This observation is explained by experimental design. Before starting the experiment, Bicarb0 and Bicarb16 required the addition of hydrochloric acid to ensure pH was 8.0. This step was necessary to guarantee postfilter pH never exceeded pH 8.0, because higher pH levels could not be read by our blood gas machine. Figure 6 demonstrates that this resulted in a reduced postfilter SID, which was because of elevated chloride concentrations (see Figure S1, Supplemental Digital Content, http://links.lww.com/ASAIO/A343). Therefore, although we removed more total CO2 with Bicarb0, compared with Bicarb16, the concomitant reduction in SID resulted in a lower final pH. Acidosis, as a result of increased plasma chloride concentration, is well described in patients receiving chloride-rich solutions.30 Because less hydrochloric acid was required to adjust the pH of Bicarb16, the effect was smaller. As a result, the pH change with Bicarb16 closely follows Prismasol, particularly at lower dialysis flow rates (Figure 5).
Carbon dioxide removal plateaued when dialysate flow rate approached plasma flow rate. Under these conditions, estimated CO2 removal was 50 ml·min−1 for Bicarb16 and 90 ml·min−1 for Bicarb0, when prefilter PCO2 was 90 mm Hg (Table 3). These removal rates compete with commercially available low-flow ECCO2R devices. The Hemolung (Alung, Pittsburgh, PA) typically removes 90 ml·min−1 of CO2,6 while the Prismalung (Baxter, Deerfield, IL) reportedly removes 60 ml·min−1 of CO2.31 Although most adults produce around 250 ml·min−1 of CO2, depending on sex, size, and metabolic state, these levels of CO2 removal are clinically beneficial when used to support ultraprotective ventilation strategies, or as an alternative to intubation in exacerbations of COPD.4,5
Importantly, we only require a blood flow rate of 214 ml·min−1 to achieve these CO2 removal rates, whereas the Hemolung and Prismalung require blood flow rates >400 ml·min−1. Unfortunately, despite the lower blood flow rate, we are delivering a high effluent dialysis dose of 171 ml·kg−1·hr−1. Conventional effluent dialysis doses are 20–40 ml·kg−1·hr−1.32 High effluent doses risk harm, such as excessive glucose and electrolyte losses.33 When our dialysis effluent dose is 69 ml·kg−1·hr−1, our CO2 removal rates are 58.8 ml·min−1 if starting PCO2 is 90 mm Hg (Bicarb16, Table 3). However, it may be possible to reduce exposure to high effluent dialysis doses by recirculating dialysate after removal of bicarbonate. The concept of dialysate regeneration is already established,34 and developing technology to remove bicarbonate alone is likely to be simpler than fully regenerating a dialysate for renal support.
Several experimental limitations should be noted. First, our modeling and in-vitro experiments have not taken into account the contribution of the red blood cell to plasma SID. When CO2 is removed from blood using conventional membrane lungs, the expected rise in pH is partially offset by a reduction in the SID.35 This occurs because red blood cells release chloride in response to the acute change in membrane potential as a result of bicarbonate losses.36 However, this effect from red blood cells would not change our prediction that bicarbonate can be removed without being replaced. It is also possible that the presence of red blood cells in the plasma may increase CO2 removal, because the presence of carbonic anhydrase in whole blood will facilitate more rapid conversion of CO2 to bicarbonate for removal. It has previously been shown that the addition of carbonic anhydrase to hollow fibers used in ECCO2R increases CO2 removal.37
Second, the absence of red blood cells prevents conclusions being drawn about the potential osmotic effects of our dialysate. But, since the main determinates of osmolality are strong ions, by aiming for a physiologic SID we can expect our dialysate to exhibit osmotic compatibility. Furthermore, a recently published study using a low bicarbonate dialysate did not report any red cell lysis in animal blood.38 Finally, our experiments are based around constructing a dialysis solution that has the same SID as plasma entering the dialysis filter. In clinical practice, this assumption is unrealistic, because patients have a range of SID depending on the underlying acid-base and electrolyte derangements associated with critical illness. These may be influenced by anything from the resuscitation fluids used, through to protein losses and nutritional deficits.39 However, our results with Bicarb16 show that pH can be restored, even if the SID of the dialysate does not exactly match that of plasma, provided it is within a physiologic range. This also seems to be the case when the SID of intravenous fluids is considered.40
Our modeling and in-vitro experiments show that bicarbonate dialysis is feasible, and that pH can be maintained without replacing removed bicarbonate, provided the SID and plasma proteins (i.e., Atot) are maintained within physiologic ranges. Estimated CO2 removal is comparable to commercially available low-flow ECCO2R devices, but dialysis dose may be excessive. Large animal experiments are now required to test scalability of our bicarbonate dialysis system and evaluate the effects on whole blood, so we can translate our findings into the broader clinical context.
The authors thank Professor Wee Han Ang from the Department of Chemistry at the National University of Singapore for his assistance with the electrolyte analysis.
1. Cove ME, MacLaren G, Federspiel WJ, Kellum JA. Bench to bedside review: Extracorporeal carbon dioxide
removal, past present and future. Crit Care 2012.16: 232.
2. Brenner K, Abrams DC, Agerstrand CL, Brodie D. Extracorporeal carbon dioxide
removal for refractory status asthmaticus: Experience in distinct exacerbation phenotypes. Perfusion 2014.29: 26–28.
3. Fanelli V, Ranieri MV, Mancebo J, et al. Feasibility and safety of low-flow extracorporeal carbon dioxide
removal to facilitate ultra-protective ventilation in patients with moderate acute respiratory distress sindrome. Crit Care 2016.20: 36.
4. Burki NK, Mani RK, Herth FJF, et al. A novel extracorporeal CO(2) removal system: Results of a pilot study of hypercapnic respiratory failure in patients with COPD. Chest 2013.143: 678–686.
5. Tiruvoipati R, Buscher H, Winearls J, et al. Early experience of a new extracorporeal carbon dioxide
removal device for acute hypercapnic respiratory failure. Crit Care Resusc 2016.18: 261–269.
6. Bermudez CA, Zaldonis D, Fan MH, Pilewski JM, Crespo MM. Prolonged use of the Hemolung respiratory assist system as a bridge to redo lung transplantation. Ann Thorac Surg 2015.100: 2330–2333.
7. Hermann A, Riss K, Schellongowski P, et al. A novel pump-driven veno-venous gas exchange system during extracorporeal CO2
-removal. Intensive Care Med 2015.41: 1773–1780.
8. Karagiannidis C, Strassmann S, Brodie D, et al. Impact of membrane lung surface area and blood flow on extracorporeal CO2 removal during severe respiratory acidosis. Intensive Care Med Exp 2017.5: 34.
9. Gattinoni L, Agostoni A, Pesenti A, et al. Treatment of acute respiratory failure with low-frequency positive-pressure ventilation and extracorporeal removal of CO2. Lancet 1980.2: 292–294.
10. Gattinoni L, Pesenti A, Mascheroni D, et al. Low-frequency positive-pressure ventilation with extracorporeal CO2 removal in severe acute respiratory failure. JAMA 1986.256: 881–886.
11. Silvester W, Bellomo R, Cole L. Epidemiology, management, and outcome of severe acute renal failure of critical illness in Australia. Crit Care Med 2001.29: 1910–1915.
12. Gatward JJ, Gibbon GJ, Wrathall G, Padkin A. Renal replacement therapy for acute renal failure: A survey of practice in adult intensive care units in the United Kingdom. Anaesthesia 2008.63: 959–966.
13. Geers C, Gros G. Carbon dioxide
transport and carbonic anhydrase in blood and muscle. Physiol Rev 2000.80: 681–715.
14. Gille JP, Bauer P, Bollaert PE, Tousseul B, Kachani-Mansour R, Munsch L. CO2
removal with hemodialysis and control of plasma oncotic pressure. ASAIO Trans 1989.35: 654–657.
15. Nolte SH, Benfer RH, Grau J. Extracorporeal CO2
removal with hemodialysis (ECBicCO2R): How to make up for the bicarbonate loss? Int J Artif Organs 1991.14: 759–764.
16. Cressoni M, Zanella A, Epp M, et al. Decreasing pulmonary ventilation through bicarbonate ultrafiltration: An experimental study. Crit Care Med 2009.37: 2612–2618.
17. Russ M, Deja M, Ott S, et al. Experimental high-volume hemofiltration with predilutional tris-hydroxymethylaminomethane for correction of low tidal volume ventilation-induced acidosis. Artif Organs 2011.35: E108–E118.
18. Kellum JA. Clinical review: Reunification of acid-base physiology. Crit Care 2005.9: 500–507.
19. Stewart PA. Modern quantitative acid-base chemistry. Can J Physiol Pharmacol 1983.61: 1444–1461.
20. Ring T, Kellum JA. Strong relationships in acid-base chemistry - modeling protons based on predictable concentrations of strong ions, total weak acid concentrations, and pCO2. PLoS One 2016.11: e0162872.
21. Stewart PA. Kellum JA, Elbers PWG. Chapter 7 - Strong ion plus carbon dioxide
plus weak acid. in Stewart’s Textbook of Acid-Base, 2009, 2nd ed. Amsterdam, The Netherlands, pp. 133–165.http://Lulu.com
22. Doberer D, Funk GC, Kirchner K, Schneeweiss B. A critique of Stewart’s approach: The chemical mechanism of dilutional acidosis. Intensive Care Med 2009.35: 2173–2180.
23. Stewart PA. Kellum JA, Elbers PWG. Chapter 6 - Strong ions plus carbon dioxide
(isolated, intact interstitial fluid). in Stewart’s Textbook of Acid-Base. 2009, Amsterdam, The Netherlands, pp. 111–132.
24. Dumler F, Levin NW. Leukopenia and hypoxemia. Unrelated effects of hemodialysis. Arch Intern Med 1979.139: 1103–1106.
25. Kolff WJ, Berk HT, Welle ter M, van der LEY AJ, van Dijk EC, van Noordwijk J. Artif Kidney, 1997:1959–1965.
26. Tolchin N, Roberts JL, Hayashi J, Lewis EJ. Metabolic consequences of high mass-transfer hemodialysis. Kidney Int 1977.11: 366–378.
27. Hakim RM, Pontzer MA, Tilton D, Lazarus JM, Gottlieb MN. Effects of acetate and bicarbonate dialysate in stable chronic dialysis patients. Kidney Int 1985.28: 535–540.
28. Leunissen KM, Hoorntje SJ, Fiers HA, Dekkers WT, Mulder AW. Acetate versus bicarbonate hemodialysis in critically ill patients. Nephron 1986.42: 146–151.
29. Desai N. Basics of base in hemodialysis solution: Dialysate buffer production, delivery and decontamination. Indian J Nephrol 2015.25: 189–193.
30. Yunos NM, Bellomo R, Story D, Kellum J. Bench-to-bedside review: Chloride in critical illness. Crit Care 2010.14: 226.
31. Godet T, Combes A, Zogheib E, Jabaudon M, Futier E, Constantin J. Novel carbon dioxide
removal device driven by a renal-replacement system without hemofilter: An experimental approach and validation Critical Care 2014.18: P316.
32. Palevsky PM, Zhang JH, O’Connor TZ, et al. Intensity of renal support in critically ill patients with acute kidney injury N Engl J Med 2008.359: 7–20.
33. Bellomo R, Cass A, Cole L, et al.; RENAL Replacement Therapy Study Investigators: Intensity of continuous renal-replacement therapy in critically ill patients. N Engl J Med 2009.361: 1627–1638.
34. Roberts M. The regenerative dialysis (REDY) sorbent system Nephrology 1998.4: 275–278.
35. Langer T, Scotti E, Carlesso E, et al. Electrolyte shifts across the artificial
lung in patients on extracorporeal membrane oxygenation: Interdependence between partial pressure of carbon dioxide
and strong ion difference. J Crit Care 2015.30: 2–6.
36. Westen EA, Prange HD. A reexamination of the mechanisms underlying the arteriovenous chloride shift. Physiol Biochem Zool 2003.76: 603–614.
37. Arazawa DT, Oh HI, Ye SH, et al. Immobilized carbonic anhydrase on hollow fiber membranes accelerates CO(2) removal from blood. J Memb Sci 2012.404–404: 25–31.
38. May AG, Sen A, Cove ME, Kellum JA, Federspiel WJ. Extracorporeal CO2 removal by hemodialysis: In vitro
model and feasibility. Intensive Care Med Exp 2017.5: 20.
39. Lee JW. Fluid and electrolyte disturbances in critically ill patients. Electrolyte Blood Press 2010.8: 72–81.
40. Morgan TJ, Venkatesh B, Hall J. Crystalloid strong ion difference determines metabolic acid-base change during in vitro
hemodilution. Crit Care Med 2002.30: 157–160.