Evaluation of the impact of extracorporeal membrane oxygenation (ECMO) oxygenator shunt flow on neonatal hemolysis: An in vitro study

Introduction

Extracorporeal membrane oxygenation (ECMO) is utilized in critically ill neonates with severe cardiopulmonary failure. ¹ Although ECMO is life-sustaining, it has its drawbacks. Complications of ECMO include stroke, thrombosis, hemolysis, hemorrhage, and infection.^1–4 Hemolysis is a considerable complication, especially in the neonatal ECMO population. Morbidity and mortality significantly increase in neonates with significant hemolysis, which include increased risk of renal failure, anemia, thrombotic complications, and bleeding.^1–5 Clinically, patients are monitored for hemolysis using non-specific markers including plasma-free hemoglobin (PFH), lactate dehydrogenase (LDH), d-dimer, and bilirubin. Of these markers, PFH is very efficacious for detecting hemolysis and centrifugal pump thrombosis. ⁶ Centrifugal pumps are used in several ECMO centers and significantly contribute to hemolytic complications.^7–10 Other factors contributing to hemolysis in neonates include the resistance from smaller cannulas and smaller tubing that cause increased shearing forces.^2,3 Additionally, ECMO circuits cause increasing inflammatory reactions, further contributing to thrombosis and coagulopathic complications.^2,5

Oxygenators are another source of hemolysis. ² Centrifugal pumps often consist of larger circuits and oxygenators intended for use on adult patients; however, size mismatches can also contribute to hemolysis, especially in the neonatal population. ² Practice variability exists among major centers for managing ECMO in neonates. Several institutions, including our own, use adult oxygenators for neonates and children because of the increased risk of thrombosis using pediatric oxygenators. ¹¹ One study that was consistent with this determined that the oxygenator was one of the most frequent sources of clot related complications in pediatric and neonatal ECMO circuits. ¹¹ With limited options for pediatric oxygenators, many centers utilize adult oxygenators with shunts. These circuits are frequently modified by adding a blood shunt from post-oxygenator to pre-oxygenator lines because centrifugal pumps often require a minimum flow rate. This shunt is regulated by a partially occluding clamp, which is meant to optimize flow to the oxygenator and the patient. We hypothesized that Hoffman clamps, which partially occlude the tubing, significantly contribute to hemolysis in the neonatal circuits. An in vitro study was conducted measuring the amount of PFH as an indicator of hemolysis in relation to occlusion by the Hoffman clamp.

Materials and methods

This in vitro study was conducted using an integrated polymethylpentene (PMP) oxygenator and centrifugal pump (Cardiohelp HLS 5.0), ¼″ arteriovenous (AV) loop tubing, and Hoffman clamps (Figure 1). A Cardiohelp HLS 5.0 with a centrifugal pump was selected because this is the standard oxygenator and pump used at our urban quaternary pediatric medical center in neonates. The Cardiohelp HLS system is manufactured by Getinge USA Sales, LLC, located at 1 Geoffrey Way, Wayne, NJ 07470, USA. The circuits were primed with 325 mL of expired ABO compatible packed red blood cells (pRBCs), 33 mL of 25% albumin, 80 mL of 5% albumin, 20 mL of 20 mEq of sodium bicarbonate, 200 units of heparin, and 1.5 mL of 150 mEq of calcium chloride. The total volume for each circuit was 461 mL. Institutional review board (IRB) approval was obtained for the study. The expired ABO-compatible blood was pooled into a cell saver device and washed. Three ECMO circuits were primed evenly with the albumin, sodium bicarbonate, heparin, and calcium chloride. The pooled blood was distributed evenly across the three circuits. A closed, soft-shelled reservoir was used to allow for circulation of the circuit without ECMO cannulas. Hematocrit was measured at 31% for each circuit, which is within neonatal physiologic limits. PFH was measured immediately after priming to standardize the baseline levels in each circuit. Shunts in these circuits are set up to go from post-oxygenator and recirculate to the pre-oxygenator tubing using 1/4″ tubing for the shunt itself and a “Y” connector to the venous ¼″ tubing limb . Our institutions oxygenators have a minimum flow rate of 500 mL/min per manufacturer recommendations. Circuits were labeled “Circuit A,” “Circuit B,” and “Circuit C.” Circuit A had a completely occluded shunt, generating 500 mL/min of arterial flow to the reservoir (Figure 2). Circuit B had a partially occluded shunt that allowed 500 mL/min of flow to be shunted back to the oxygenator (Figure 3). The partially occluding clamp was tightened to allow 500 mL/min of flow to the reservoir and 500 mL/min through the shunt. Circuit C had a fully open shunt, generating 500 mL/min of arterial flow to the reservoir and generating 1000 mL/min of flow back to the oxygenator (Figure 4). A flow probe was placed on the arterial limb to confirm arterial flow to the reservoir. The arterial flow for each circuit was kept constant at 500 mL/min, which was measured by the arterial probe. Another flow probe was placed on the venous limb, after insertion of the shunt, which measured flow from the reservoir in addition to the shunt. The venous limb flow probe thereby can be used as a surrogate for total flow through the oxygenator. The venous flow, which was measured immediately pre-oxygenator and after insertion of the shunt (if applicable), was 500 mL/min, 1000 mL/min, and 1500 mL/min in circuit A, circuit B, and circuit C respectively. The revolutions per minute (RPM) in circuit A, circuit B, and circuit C were 1530, 1675, and 1405, respectively. The pressure (P_Int) measured pre-oxygenator, and post-centrifugal pump was 97 mmHg, 107 mmHg, and 79 mmHg in circuit A, circuit B, and circuit C respectively. The pressure (P_Art), which is measured post-oxygenator, was 96 mmHg, 107 mmHg, and 78 mmHg in circuit A, circuit B, and circuit C respectively. After each circuit was primed, the blood was oxygenated. Baseline PFH was measured from each circuit and again at serial intervals for 5 days. PFH was measured using a Thermo Scientific Evolution 201 equipment, which is an ultraviolet visible spectrophotometer. The software for the spectrophotometer is Thermo Scientific InSight application.

OPEN IN VIEWER

Figure 1.

Hoffman clamp.

OPEN IN VIEWER

Figure 2.

Circuit A-fully occluded shunt with 500 mL/min of arterial flow to the reservoir and 500 mL/min of venous flow back to the oxygenator.

OPEN IN VIEWER

Figure 3.

Circuit B-partially occluded shunt with 500 mL/min of arterial flow to the reservoir, 500 mL/min of flow through shunt equaling 1000 mL/min of venous flow through oxygenator.

OPEN IN VIEWER

Figure 4.

Circuit C-fully open shunt with 500 mL/min of arterial flow to the reservoir, 1000 mL/min of flow through shunt equaling 1500 mL/min of venous flow through oxygenator.

Results

The baseline PFH and the average rate of increase for each circuit were tabulated (Table 1). A PFH versus time graph was obtained (Figure 5). Circuit C, which had the fully open shunt and more blood being redirected to the oxygenator, showed the greatest increase in PFH throughout the experiment (26.3 mg/dL/day). Circuit A (fully occluded shunt) and Circuit B (partially occluded shunt) had a PFH rate increase of 14.1 mg/dL/day and 12.9 mg/dL/day, respectively, suggesting a negligible difference.

OPEN IN VIEWER

Table 1.

Evaluation of plasma-free hemoglobin levels between circuits.

	Circuit A	Circuit B	Circuit C
Time (hours)	Fully occluded shunt (mg/dL)	Partially occluded shunt (mg/dL)	Open shunt (mg/dL)
0 (baseline)	25.4	26	24.8
5	24.8	24	23.6
23	27.4	27.4	30.2
28	28.7	29	32.6
46	31.6	31.6	41.4
52	34.1	33.2	46
69	49.6	43.6	66.5
76	55.5	47.1	76.4
93	80.1	75.8	126.6
Average rate of hemolysis (mg/dL/day)	14.1	12.9	26.3

OPEN IN VIEWER

Figure 5.

Plasma-free hemoglobin (mg/dL) versus time (hours).

Discussion

This in vitro study showed that partially occluding Hoffman clamps, which regulate shunt flow in centrifugal ECMO circuits, do not significantly contribute to hemolysis. The highest hemolysis rate was seen in Circuit C, which had a fully open shunt. The reason for Circuit C having the highest rate of PFH increase is likely because more blood was shunted back through the oxygenator. This suggests that the Hoffman clamps are not significant contributors to hemolysis, but perhaps the oxygenator may contribute more significantly to hemolysis. Circuit A and Circuit B rates of hemolysis were not significantly different. Although Circuit B did have flow back to the oxygenator as well, it was only 500 mL/min (compared to 1000 mL/min in Circuit C), which suggests that after a certain threshold, the rate of hemolysis increases as more blood flows back into the oxygenator and pump through the shunt. In this study, the threshold of flow that caused in increase in hemolysis was likely somewhere between 1000 mL/min and 1500 mL/min as noted by the venous flows in circuit B and circuit C. The current practice at our institution for patients under 19 kg is to use ¼″ tubing in the ECMO circuit with the Cardiohelp oxygenator, which requires a shunt. Our shunts are limited from 400 to 600 mL/min to meet factory minimum oxygenator flow rate, which is supported with the findings of this study.

The data from these circuits also suggests that having a shunt and partially occluding clamps directing flow from the centrifugal pump may even be a protective factor for hemolysis. This finding is significant because practice variability exists among different ECMO centers. Some centers keep the shunt fully open without using a partially occluding clamp to regulate flow. Further investigation may help elucidate if keeping a shunt fully open can contribute to increased hemolysis compared to partially or fully occluding the shunt.

The circuit configuration is essential to limiting hemolysis. It has been illustrated that larger circuits in smaller patients (neonates) can significantly contribute to hemolysis due to increased surface area for shear forces to occur. ² Given the increased morbidity and mortality associated with increased rates of hemolysis, efforts to limit it is essential to improve care of critically ill neonates on ECMO. Lower flow rates have also been associated with increased hemolysis in centrifugal pumps.^2,9 It has been demonstrated that pump efficiency drops with flows of 500 mL/min compared to 4 L/min². It is common for neonates to have significantly lower flow rates, especially under 1 L/min. This further emphasizes the need to limit hemolysis in the circuit, including evaluating shunts and partially occluding clamps in vivo.

Loop configuration components including connectors, restrictive valves, and even loop angulation have been suspected sources of hemolysis and thrombogenicity in vitro, which stresses the importance of keeping the same loop configurations across circuits during testing.^12,13 The suspected mechanism of loop components causing hemolysis is platelet activation and back pressure into centrifugal pump heads, especially in VAD circuits that only have one outlet. ¹³ In our configuration, the circuits and loop configurations were identical except for the degree of clamping on the shunt. The difference between our Cardiohelp centrifugal pump configuration and a VAD is that post-oxygenator there are two outlets, the shunt and arterial flow, which thereby reduces back pressure onto the centrifugal pump and integrated oxygenator.

A limitation of this experiment is in its in vitro design; therefore, it does not account for patient factors in vivo, which limits the applicability to inpatient neonates. Although the clinical applicability is limited, this model controls other aspects of hemolysis so that the hypothesis can be reliably tested. Another limiting factor of the study in applying this clinically is lack of cannulas in the circuit configuration, which likely contribute to hemolysis in vivo considering the resistance provided by neonatal sized cannulas. Neonatal cannulas are significantly smaller and, therefore, cause increased hemolysis compared to larger cannulas, which is not accounted for in this study. In this model, it was advantageous not to use cannulas to eliminate other contributors to hemolysis to isolate the effects of the occluding clamp. The aim of this study was to evaluate if hemolysis is caused by Hoffman clamps; therefore, the study did not account for hemolysis caused by the pump itself. The centrifugal pump was the same brand and configuration across all three circuits so we are presuming that they cause the same rate of hemolysis at the same RPM. Whole blood was not used for the experiment. Derangements in other blood cell lines and the coagulation pathway, including International Normalized Ratio (INR) and platelets, can contribute to hemolysis in neonates; therefore, use of only pRBCs in this experimental set up could have affected the results and applicability to animal and human models. ⁴ Finally, another limitation is using three circuits ran over 5 days, and the experiment was not repeated to show its reproducibility. Although it is difficult to draw strong conclusions from a single run, our data suggests is consistent with increased flow through the oxygenator is a potential cause for increased hemolysis.

This experimental data suggests that the partially occluding clamps that regulate ECMO shunt flow were not significant causes of hemolysis. From our data, it appears that increased flow through the shunt increases flow through the oxygenator, thereby generating more hemolysis. Our circuits were set up in three configurations that would be used clinically for a neonate with constant arterial flow; which is a surrogate for flow to the patient. It is known that oxygenators cause hemolysis so our data suggests that the reason for increased hemolysis in the open shunt circuit was due to increasing total flow through the oxygenator. ² The RPM, pre-oxygenator pressure, delta pressure, and post-oxygenator pressures were comparable, and the circuit tubing length and radius were constant across all three circuits. All of these factors may contribute to hemolysis, so we tried to control them as best as possible. Additionally, the blood was washed and pooled so that factors like viscosity of fluid were consistent among the three circuits. The findings of this in vitro study suggest that using a shunt and partially occluding clamp may limit the degree of hemolysis compared to a fully open shunt and may have clinical applicability in how ECMO circuits are catered to the neonatal population. Important to note, the scope of this experiment was to test impact of hemolysis from the Hoffman clamp, not the centrifugal pump. Further studies reproducing these findings, using whole blood, cannulas, animal models, and eventually humans, would be beneficial to assess the significant contributors to hemolysis in neonatal ECMO circuits. Other future studies aim to evaluate the impact of hemolysis in neonates from using pediatric and neonatal specific oxygenators.