Mechanical ventilation (MV) is widely used as the standard of care for respiratory failure; however, long-term use leads to ventilator-induced injury. Patients with acute respiratory disease syndrome (ARDS) are often treated using extracorporeal membrane oxygenation (ECMO). ECMO has been considered a “bridge to nowhere” as long-term use requires patients being confined to an ICU bed which can increase morbidity and mortality.1,2 Patients also experience progressive muscle deconditioning which can further increase morbidity. To mitigate morbidity, clinicians have introduced ambulation into clinical practice.3–5 Ambulating ECMO patients has improved treatment outcomes.6–8 Respiratory support including the Quadrox oxygenators or Maquet CardioHelp used in conjunction with the Avalon Elite Dual Lumen Cannula have facilitated ambulation. Contemporary ECMO systems can simplify ambulation, if they are made more compact. The work in this article addresses some key limitations of contemporary ECMO.
Work has been focused on developing artificial lung devices that can potentially ambulate patients. The University of Maryland has produced the Ambulatory Pump Lung (APL), and our group is developing the Paracorporeal Ambulatory Assist Lung (PAAL). These are integrated and compact blood pump-oxygenators for respiratory assist. The APL device has been successful in chronic animals.9 Our group has reported a hollow fiber membrane (HFM) bundle design that has an efficiency (oxygenation per unit surface area) over 275 mL/min/m2 at 3.5 L/min blood flow.10 We are developing the PAAL using this HFM bundle and have reported successful acute and chronic animal studies.11,12 The APL and PAAL can potentially be used as wearable devices capable of ambulating patients and providing a viable bridge to lung transplant or recovery.
Not all patients require the full respiratory support ECMO provides. Patients with acute exacerbations of chronic obstructive pulmonary disease (COPD) or mild to moderate ARDS benefit from partial CO2 removal (CO2R).13 For more severe cases of ARDS, oxygenation support is required and extracorporeal CO2 removal (ECCO2R) may be used to allow for lung protective ventilation.14,15 ECCO2R works similarly to ECMO, but at blood flow rates below 1 L/min and is intended to remove CO2. At these blood flow rates, oxygenation support is not provided; however, the lower blood flow rates permit less invasive cannulation. The Hemolung RAS has been clinically used for ECCO2R to correct hypercapnia in COPD and ARDS patients at blood flow rates up to 500 ml/min.16–18 Compact ECCO2R systems would further simplify in-hospital ambulation. The PAAL could be used as a wearable, pump-driven ECCO2R device.19
A smaller sized device can improve portability and allow for ambulation. Device size can be decreased by increasing gas exchange efficiency (gas exchange per unit surface area). Gas exchange efficiency in artificial lungs is limited by the fluid boundary layer that is formed at the fiber surface. This boundary layer thickness scales approximately with the square root of velocity.20 Blood recirculation can increase velocity through the HFM bundle. Patent literature has suggested the use of blood recirculation21,22; however, recirculation has not been computationally or experimentally demonstrated as a practical means for designing artificial lungs. This study investigates the effect of blood recirculation on gas exchange in the PAAL.
In this study, we targeted 180 ml/min oxygenation at 3.5 L/min, CO2R of 70 ml/min at 500 ml/min, surface area under 0.65 m2, and hemolysis due to recirculation < 0.05 gm/100 L. A mass transfer model predicted the effect of blood recirculation on oxygenation and CO2R. Using the model, we selected a fiber surface area that would meet our oxygenation, or CO2R, requirements. A computational fluid dynamics (CFD) guided approach was used to analyze the hydrodynamics and shear stresses. We then measured in-vitro hemolysis and gas transfer in a proof of concept prototype.
A previously described oxygen transfer model10,11 predicted oxygenation in the HFM bundle. The fiber diameter was set as the hydraulic diameter, and interstitial velocity was selected for calculating the Reynold’s number. Recirculation was simulated by iteratively calculating the oxygenation rate while the inlet saturation () changed. was determined as follows: where is the device outflow, is the recirculation flow rate, is the outlet oxygen saturation, and is the constant inlet saturation (65%). The recirculation flow rate represents the flow rate in the recirculation limb. The iteration was stopped when changed less than 1%. Oxygenation from recirculation was simulated while varying surface area between 0.2 and 0.8 m2 and maintaining constant flows through the device (, ). The intended application of the PAAL is in a veno-venous configuration; thus, was set to 3.5 L/min. Based on feasibility bench testing, the maximum recirculation rate achievable was 6.5 L/min, thus the choice of . The surface area meeting the oxygenation target (180 ml/min) was chosen for further evaluation.
Similarly, CO2R was modeled by a previously described gas transfer model.19,23 The inlet pCO2 was calculated as: where is the outlet pCO2 and is the constant inlet pCO2 (45 mm Hg). The CO2R rate was solved iteratively, and was changed at each iteration until changed less than 0.01%. CO2R results were simulated for HFM surface areas between 0.1 and 1 m2. The smallest surface area that met the CO2R requirement was selected.
A proof of concept prototype was constructed for evaluating gas exchange before investing in the fabrication of the simulated design. The surface area of the proof of concept prototype was based on the mathematical model and was constructed from stacked sheets of polymethylpentene hollow fibers (44 fibers/inch) using previously described methods.10 This bundle was assembled into a previously described PAAL prototype.11 A recirculation channel external to the PAAL (Figure 1) was formed using two Y-connectors (Qosina, Ronkonoma, NY) and R-3603 Tygon tubing (Fisher Scientific, Hampton, NH).
Similarly, a 0.3 m2 polymethylpentene HFM bundle was fabricated for CO2R characterization. Our Pittsburgh Pediatric Ambulatory Lung (P-PAL) device uses a 0.3 m2 HFM bundle and was used as the test fixture for the ECCO2R experiments.24 The recirculation channel was created as in the oxygenation experiment. The impeller was removed from the P-PAL and a Centrimag pump (Abbott Laboratories, Abbott Park, IL) was placed before the P-PAL to drive flow. The Centrimag was used to pump in this system to evaluate the effect of recirculation on CO2R before investing in an integrated recirculation device.
In-Vitro Gas Exchange
Gas transfer was characterized using bovine blood in a previously described single pass system.10,11 The loop comprised the prototype, a downstream “deoxygenator”, and an upstream Centrimag pump (Abbott Laboratories). Components were connected in series with 6 L compliant reservoir bags. For oxygenation, the device outflow was maintained constant at 3.5 L/min and measured using a Centrimag flow probe and controller. Recirculation flow rate was varied (0, 2.5, 4.5, and 6.5 L/min) using a Hoffman’s clamp to adjust the resistance of the recirculation channel. A 1 ml blood sample from the inlet and outlet were drawn after a circulation time of 45 seconds. For CO2R, the device outflow was set to 500 ml/min and measured using a Transonic flow probe (Transonic Systems Inc., Ithaca, NY). Recirculation flow rate (0, 2.5, 3, and 5.5 L/min) was varied as previously described. The inlet pCO2 was set to either 45 ± 5 mm Hg (normocapnic condition) or 75 ± 5 mm Hg (hypercapnic condition). The outlet sweep gas concentration was measured using a WMA-4 Gas Analyzer (PP System, Amesbury, MA). A 1 ml blood sample was taken from the inlet and outlet of the device once the outlet sweep gas CO2 concentration changed by less than 10 ppm. The pure oxygen sweep gas flow rate was varied between 9 and 14 L/min. Preliminary trials were conducted to ensure the device was not sweep gas limited. A RapidPoint 405 blood gas analyzer (Siemens, Munich, Germany) was used to measure oxygen saturation, hemoglobin, and gas partial pressures. Each data point was repeated three times. Gas exchange data were statistically analyzed using an analysis of variance (ANOVA). The analysis was followed with a Bonferroni post hoc. All analyses were conducted in IBM SPSS Statistics 24 (Armonk, NY).
We used a CFD-guided approach for incorporating recirculation in the PAAL at the required surface area. Computational fluid dynamics was used to analyze the hydraulic and hemodynamic aspects of the PAAL device. Blood flow velocities and pressures within the PAAL were modeled via laminar CFD analysis performed using ANSYS Fluent v17 (ANSYS, Canonsburg, PA). Blood was treated as a homogeneous incompressible fluid of density 988 kg/m3 and constant viscosity of 3.4 cP. The fiber bundle was modeled as porous media with a uniform viscous resistance of 1e9 1/m2 and a fluid porosity of 0.58.23 Each CFD mesh consisted of approximately 1.5M tetrahedral cells, and rotor motion was handled using a frozen relative motion frame of reference.
The hemodynamic goals were to achieve 250 mm Hg pressure rise, a nominal device outflow of 3.5 L/min, and a recirculation flow through the bundle of 10.0 L/min, while minimizing shear stresses in the device. Four variants of recirculation loop geometry were evaluated, two of which incorporated stenotic nozzles. Hydrodynamic performance of this design was computationally determined at 2,000, 2,250 and 2,500 RPM. Recirculation flow rate, pressure generated, and max shear at each data point were determined. The geometry reported herein best met the hemodynamic goals as determined from CFD analyses only.
Hemolysis was measured for 6 hours in a test and a control loop. Each loop comprised the oxygenation prototype connected to an 800 ml reservoir. The test loop contained the experimental prototype (Figure 1), and the control loop comprised the same prototype without an HFM bundle. Plasma-free hemoglobin (PfHb) was measured every 30 minutes for the first 2 hours and then every hour. The rise in PfHb over time was quantified as a normalized index of hemolysis (NIH). Device outflow was constant at 3.5 L/min, and the recirculation flow rate was 6.5 L/min. Procedural details for in-vitro testing are described elsewhere.10
The model predicted a bundle surface area of 0.4 m2 met oxygenation requirements (180 ml/min) at a recirculation rate of 6.5 L/min (Figure 2A). The predicted oxygenation enhancement from recirculation was 20%. The mass transfer model predicted that a bundle surface area of 0.3 m2 met the CO2R requirement (70 ml/min) at a recirculation flow rate of 2.5 L/min (Figure 2B).
In-Vitro Gas Exchange
In-vitro oxygenation was 180 ml/min ± 8 (efficiency exceeding 470 ml/min/m2) at a recirculation flow rate of 6.5 L/min and device outflow of 3.5 L/min (Figure 3A). Recirculation significantly increased oxygenation at 4.3 L/min (p = 0.042) and at 6.5 L/min (p = 0.026) relative to no recirculation. Oxygenation was predicted within ± 8%. In-vitro CO2R at a recirculation flow rate of 5.5 L/min was 61.9 ml/min under normocapnic conditions, a 10.4% increase. The model over predicted by up to 32%. Under hypercapnic conditions, CO2R was 90.0 ml/min at a recirculation flow rate of 5.5 L/min, a 7.2% increase. The model over predicted hypercapnic results by up to 16% (Figure 3B).
Hydrodynamic performance from the CFD analysis shows the pressure generated increases from 190 mm Hg at 2,000 RPM to 330 mm Hg at 2,500 RPM and recirculation flow rate increases from 5.1 L/min at 2,000 RPM to 6.9 L/min at 2,500 RPM (Figure 4). The intersection of these curves with the Avalon DLC cannula curve indicates that the operating point is between 2,250 and 2,500 RPM. All simulated points had shear stresses under 300 Pa and uniform flow through the bundle. Figure 5 shows the shear stress map and flow velocity vectors at the device’s intended operating point (2,500 RPM, device outflow of 3.5 L/min, and recirculation flow of 6.5 L/min).
The NIH of the prototype with the bundle was 0.102 gm/100 L, and the NIH without the bundle was 0.09 gm/100 L. The contribution of just the bundle was 0.012 gm/100 L (Figure 6).
Ambulating patients on ECMO improves treatment outcomes.7,8 We are developing the PAAL for simplified ambulation. In this study, we tested blood recirculation as means for increasing gas transfer efficiency and reducing device size. We developed a recirculating flow design using a modeling approach. Computational fluid dynamics analysis indicated that the proposed design can produce 3.5 L/min at 250 mm Hg (overcomes the Avalon 27 Fr. DLC resistance) while maintaining shear stresses under 300 Pa. Our oxygenation model predicted the experiment within 6%. In-vitro oxygenation on our prototype reached 180 ml/min ± 8 at 3.5 L/min of device outflow and 6.5 L/min of recirculating flow. The maximum predicted oxygenation enhancement was 20%. In-vitro hemolysis was acceptable at these flows. In-vitro low-flow CO2R reached 62 ml/min (normocapnic) and 90 ml/min (hypercapnic) at 500 ml/min of device outflow and 5.5 L/min of recirculating flow. Recirculation resulted in a maximum in-vitro CO2R enhancement of 10%, whereas the predicted CO2R enhancement was up to 46%. Our gas transfer model over predicted in-vitro results by up to 32%. We met the oxygenation target with a 0.4 m2 HFM area using the recirculation approach (3.4 times smaller than the clinical standard).25 We did not meet the CO2R target (70 ml/min at normocapnia) at an HFM bundle surface area of 0.3 m2.
While patent literature suggests recirculation can enhance gas transfer efficiency,16,17 our study is the first evaluation of recirculation in an artificial lung device. Oxygenation efficiency (oxygenation rate per membrane surface area) at 3.5 L/min in the recirculating design was 1.7 to 4.9 times higher than existing artificial lung devices (Table 1). The APL has a 0.8 m2 annular HFM bundle integrated with a magnetically levitated centrifugal pump.26 The Thoracic Artificial Lung (TAL) has a 2.4 m2 HFM bundle and uses the native heart as a pump.27 The PAAL we previously reported has a 0.65 m2 stacked HFM bundle and is integrated with a magnetically driven centrifugal pump.11 Oxygenation in our proposed recirculation design compares to these devices at 3.5 L/min, yet has a lower membrane surface area (0.4 m2). Low flow CO2R efficiency under normocapnic conditions in the recirculating design ranged from 186 to 206 ml/min/m2. Despite a 10% increase in enhancement, the targeted CO2R rate was not met by the recirculation device.
Blood flowing through the HFM bundle forms a boundary layer at the fiber surface. Gas transfer in HFM devices is limited by the thickness of this layer which scales approximately with the square root of velocity.15 Our previous study increased mean bundle flow velocity by reducing the frontal area of the HFM.10 We used an HFM bundle with an identical frontal area in this study. Recirculation increased the mean velocity through the oxygenation bundle by approximately three times the mean velocity in our previous study. Oxygenation efficiency enhancement in our design is from increased velocities rather than increased contact time with the HFM bundle. Our model can be modified for discerning the effects of velocity and contact time during recirculation. The mass transfer coefficient of the zero recirculation case was applied to calculate oxygenation for the 6.5 L/min recirculation case. The oxygenation for the modified model was calculated as 115 ml/min, which is 65% of the oxygenation predicted by the model with no modifications. This result indicates that oxygenation enhanced by increasing mean bundle velocity during recirculation.
In the CO2R studies, recirculation increased the superficial blood velocity through the bundle up to 11 times compared with the no recirculation case. CO2R, however, only increased 10% at normocapnic conditions and 7% at hypercapnic conditions. These marginal improvements in CO2R at both normocapnic and hypercapnic conditions suggest that the CO2 driving gradient is reduced to the point that additional gas transfer does not occur as the boundary layer is decreased. Although the recirculation stream increases the velocity of the blood through the bundle, this approach inherently decreases the CO2 concentration gradient across the fibers. The mixing of the recycle stream, which has the same pCO2 as the device outlet, with the fresh feed results in a mixed stream pCO2 approximately equal to the weighted average of the two streams. Thus, the pCO2 of the mixed stream is more similar (no more than 5.5 mm Hg higher) to the device outlet than the fresh feed. Hence, an increase in the recycle ratio did not result in an increase in CO2R. Hemolysis was not evaluated for the CO2R device because enhancement did not achieve the gas exchange target.
Interestingly, the gas exchange model over predicted the CO2R rate when blood was recirculated more so than when blood was not recirculated. An important difference between the recirculated and nonrecirculated systems is the oxygen saturation of the device inlet blood. The blood entering the device during recirculation is > 99% saturated with oxygen at all recycle ratios whereas the fresh feed is at venous conditions. The model’s empirical fit of the CO2 dissociation curve does not consider the effect of oxygen saturation has on the total CO2 concentration. The Haldane Effect describes a decrease in whole blood CO2 content as oxygen saturation increases for a given pCO2. This is a result of a decrease in carbamino hemoglobin as hemoglobin is oxygenated.28 Thus, for the recirculation cases, the CO2 dissociation curve shifts downward compared with the nonrecirculation cases; however, this shift is not represented in the model. This unaccounted for shift is a possible reason for the model’s overestimation of the effect of blood recirculation on CO2R.
The CFD modeled design showed low shear stresses (< 300Pa) which have previously correlated with low hemolysis.29 The in-vitro prototype tested in this study did not have an integrated recirculation channel. A prototype was constructed in this manner for testing the merits of recirculation on gas exchange before fabricating the CFD modeled design. Hemolysis was thus measured in the in-vitro prototype with and without the HFM bundle in place as a means for determining blood damage associated with recirculating blood through the bundle. An acceptable (< 0.05 gm/100 L)30 increase in the NIH from recirculating blood through the HFM bundle was measured at oxygenation conditions. This modeling and in-vitro work indicate that the CFD modeled design will have acceptable hemolysis.
In conclusion, a model-based approach was used for designing a device incorporating blood recirculation in this study. We then validated the model by in-vitro testing a proof of concept prototype. Our oxygenation target (180 ml/min at 3.5 L/min) was met with an efficiency over 470 ml/min/m2. Hemolysis from recirculating flows through the HFM bundle was low (0.012 gm/100 L). Recirculation is an effective means of enhancing oxygenation efficiency but did not show the same enhancement for low-flow CO2R. The proposed design for an artificial lung significantly increases oxygenation performance while maintaining low levels of in-vitro hemolysis.
The authors would like to acknowledge Dr James F. Antaki’s contribution of the fundamental concept of blood recirculation.
1. Biscotti M, Sonett J, Bacchetta M. ECMO as bridge to lung transplant. Thorac Surg Clin 2015.25: 17–25.
2. Maury G, Langer D, Verleden G, et al. Skeletal muscle force and functional exercise tolerance before and after lung transplantation: A cohort study. Am J Transplant 2008.8: 1275–1281.
3. Biscotti M, Bacchetta M. The “sport model”: Extracorporeal membrane oxygenation using the subclavian artery. Ann Thorac Surg 2014.98: 1487–1489.
4. Reeb J, Olland A, Renaud S, et al. Vascular access for extracorporeal life support: Tips and tricks. J Thorac Dis 2016.8(suppl 4): S353–S363.
5. Rajagopal K, Hoeper MM. State of the art: Bridging to lung transplantation using artificial organ support technologies. J Heart Lung Transplant 2016.35: 1385–1398.
6. Perme CS, Southard RE, Joyce DL, Noon GP, Loebe M. Early mobilization of LVAD recipients who require prolonged mechanical ventilation. Tex Heart Inst J 2006.33: 130–133.
7. Pruijsten R, van Thiel R, Hool S, Saeijs M, Verbiest M, Reis Miranda D. Mobilization of patients on venovenous extracorporeal membrane oxygenation support using an ECMO helmet. Intensive Care Med 2014.40: 1595–1597.
8. Biscotti M, Gannon WD, Agerstrand C, et al. Awake extracorporeal membrane oxygenation as bridge to lung transplantation: A 9-year experience. Ann Thorac Surg 2017.104: 412–419.
9. Wu ZJ, Zhang T, Bianchi G, et al. Thirty-day in-vivo performance of a wearable artificial pump-lung for ambulatory respiratory support. Ann Thorac Surg 2012.93: 274–281.
10. Madhani SP, Frankowski BJ, Federspiel WJ. Fiber bundle design for an integrated wearable artificial lung. ASAIO J 2017.63: 631–636.
11. Madhani SP, Frankowski BJ, Burgreen GW, et al. In vitro
and in vivo
evaluation of a novel integrated wearable artificial lung. J Heart Lung Transplant 2017.36: 806–811.
12. Madhani S, Frankowski B, YE S, et al. In vivo 5 day animal studies of a compact, wearable pumping artificial lung. ASAIO J 2017.65: 94–100.
13. Abrams D, Brenner K, Burkart K, et al. Pilot study of extracorporeal carbon dioxide removal to facilitate extubation and ambulation
in exacerbations of chronic obstructive pulmonary disease. Ann Am Thorac Soc 2013.10: 307–314.
14. Bein T, Weber-Carstens S, Goldmann A, et al. Lower tidal volume strategy (~3 ml/kg) combined with extracorporeal CO2 removal versus “conventional” protective ventilation (6 ml/kg) in severe ARDS. Intensive Care Med 2013.39: 847–856.
15. Terragni PP, Del Sorbo L, Mascia L, et al. Tidal volume lower than 6 ml/kg enhances lung protection: Role of extracorporeal carbon dioxide removal. Anesthesiology 2009.111: 826–835.
16. Akkanti B, Rajagopal K, Patel K, et al. Low-flow extracorporeal carbon dioxide removal using the hemolung respiratory dialysis system® to facilitate lung-protective mechanical ventilation in acute respiratory distress syndrome. J Extracorpor Technol 2017.49: 112–114.
17. Bonin F, Sommerwerck U, Lund LW, Teschler H. Avoidance of intubation during acute exacerbation of chronic obstructive pulmonary disease for a lung transplant candidate using extracorporeal carbon dioxide removal with the hemolung. J Thorac Cardiovasc Surg 2013.145: e43–e44.
18. Parilla F, Bergesio L, Aguirre-Bermeo H, et al. Ultra-low tidal volumes and extracorporeal carbon dioxide removal (hemolung® RAS) in ards patients. A clinical feasibility study. Intensive Care Med Exp 2015.3: A7.
19. May AG, Jeffries RG, Frankowski BJ, Burgreen GW, Federspiel WJ. Bench validation of a compact low-flow CO2
removal device. Intensive Care Med Exp 2018.6: 34.
20. Federspiel W, Henchir K. Lung, artificial: Basic principles and current applications. Encycl Biomater Biomed Eng 2008.9: 910–921.
21. Hubbard L, Clausen E. Pump/oxygenator with blood recirculation
. US Patent US5411706, 1994.
22. Wu Z, Griffith B. Blood oump-oxygenator system. US Patent Application US20070249888 A1. 2007.
23. Svitek RG, Federspiel WJ. A mathematical model to predict CO2 removal in hollow fiber membrane oxygenators. Ann Biomed Eng 2008.36: 992–1003.
24. Orizondo RA, May AG, Madhani SP, et al. In vitro
characterization of the Pittsburgh pediatric ambulatory lung. ASAIO J 2018.64: 806–811.
25. Maquet Getinge Group. HLS set advanced 2015.
26. Zhang T, Wei X, Bianchi G, et al. A novel wearable pump-lung device: In vitro
and acute in vivo
study. J Heart Lung Transplant 2012.31: 101–105.
27. Schewe RE, Khanafer KM, Arab A, Mitchell JA, Skoog DJ, Cook KE. Design and in vitro
assessment of an improved, low-resistance compliant thoracic artificial lung. ASAIO J 2012.58: 583–589.
28. Arthurs G, Sudhakar M. Carbon dioxide transport. Crit Care Pain 2005.5: 207–210.
29. Leverett LB, Hellums JD, Alfrey CP, Lynch EC. Red blood cell damage by shear stress. Biophys J 1972.12: 257–273.
30. Kawahito S, Maeda T, Yoshikawa M, et al. Blood trauma induced by clinically accepted oxygenators. ASAIO J 2001.47: 492–495.