Secondary Logo

Journal Logo

Basic Science Aspects

The Effects of Vasoconstriction And Volume Expansion on Veno-Arterial ECMO Flow

Moller, Per W.∗,†; Hana, Anisa; Heinisch, Paul Philipp∗,‡; Liu, Shengchen; Djafarzadeh, Siamak; Haenggi, Matthias; Bloch, Andreas; Takala, Jukka; Jakob, Stephan M.; Berger, David

Author Information
doi: 10.1097/SHK.0000000000001197



Extracorporeal membrane oxygenation (ECMO) has gained widespread use over the last decade for the treatment of severe cardiorespiratory failure and cardiac arrest. Despite remarkable technical improvements with improved system biocompatibility and centrifugal instead of roller pumps, the outcome of cardiogenic shock, extracorporeal resuscitation, and postcardiotomy shock remains modest at best (6–8). The physiology of ECMO is incompletely understood and data on hemodynamic support to optimize ECMO-flow with volume or vasopressors for patients on veno-arterial ECMO are scant (6,8). Low blood flow and positive fluid balance on ECMO are strong predictors of mortality (28,33). Volume expansion is the common choice for increasing ECMO flow (2), despite the risks of progressively positive fluid balance with worsening prognosis.

We and others have shown that the ECMO blood flow is directly dependent on venous return (VR) (13,22), which is described as 

The mean systemic filling pressure (MSFP) is the elastic recoil pressure in the systemic vasculature (3), caused by the stressed blood volume and the vascular compliance. It can be measured in no-flow states (27). Conceptually, venous return driving pressure (VRdP), the gradient between MSFP and right atrial pressure (RAP), drives VR against the resistance to venous return (RVR) which reflects the lumped resistance of all vascular beds for blood returning to the heart (22). As around 70% of blood resides in the venous pool and the majority is unstressed, vasoconstriction can increase stressed volume and MSFP via recruitment of unstressed volume (19). This has been shown for epinephrine (5,21), various α1- and α2-agonists (1) and norepinephrine (11). An approach to limit volume expansion while achieving increased ECMO flow may clinically be beneficial. As outcome data on optimal supportive hemodynamic treatment for VA-ECMO is currently lacking, a study on the underlying mechanisms of action for available treatments could provide the basis for clinical decision-making.

In this porcine model of ventricular fibrillation and veno-arterial (VA) ECMO support, we compared the effects of volume expansion with Ringer's lactate and vasoconstriction using norepinephrine on ECMO blood flow and delivery of oxygen (DO2). Based on the concepts of venous return and recruitment or expansion of stressed vascular volume (19), we hypothesized that both volume expansion and vasoconstriction with norepinephrine could increase maximum achievable ECMO flow, although acting on stressed vascular volume via different mechanisms. We further hypothesized that the limits of ECMO flow are set by stressed volume and vascular circuit properties, rather than performance of the mechanical pump.


The study complied with the Guide for the Care and Use of Laboratory Animals (National Academy of Sciences, 1996) and Swiss National Guidelines, and was approved by the Commission of Animal Experimentation of Canton Bern, Switzerland (BE 16/17). Twelve domestic pigs (7 female, 5 male, mean body weight 40.0 ± 2.0 kg at 12 weeks of age) were fasted for 12 h with free access to water after a 3-day quarantine under veterinary observation at the animal hospital of the University of Bern. The first three animals were used in pilot studies to establish the instrumentation and feasibility of procedures. As previously described (4,22), after premedication with intramuscular ketamine (20 mg/kg) and xylazine (2 mg/kg), vascular access was established and anaesthesia was induced with midazolam (0.5 mg/kg) and atropine (0.02 mg/kg) followed by intubation and placement of a gastric tube.

Anaesthesia was maintained with propofol (4 mg/kg/h) and fentanyl (5 μg/kg/h) and the depth was controlled by repeatedly testing the response to nose pinch and targeting a bispectral index <60 (BIS Quatro, Covidien, Mansfield, Mass). During surgery, propofol and fentanyl infusions were increased to 6 mg and 20 μg/kg/h, respectively. Additional injections of fentanyl (50 μg) or midazolam (5 mg) were given as needed. Cefuroxime (1.5 g) was given at skin incision and repeated after 4 h. Intermittent muscle relaxation was induced with rocuronium (0.5–1 mg/kg) for the study measurements. The pigs were mechanically ventilated in a volume-controlled mode (Servo-I, Maquet Critical Care, Solna, Sweden) using a PEEP of 5 cm H2O, fraction of inspired oxygen of 0.30, I:E ratio 1:2 and tidal volume 7 mL/kg body weight. Respiratory rate was adjusted to maintain an end-tidal Pco2 of 40 mmHg.


The following catheters were surgically placed: a left carotid artery catheter, a right jugular three-lumen catheter, and an introducer sheath in the right femoral vein for rapid volume exchange. Cystostomy was performed for urinary output monitoring. The thoracic cavity was entered via a median sternotomy and the pericardium was opened. After administration of 5,000 U of heparin, the right atrium (RA), the ascending aorta and left atrium were cannulated (29 Fr 3-stage venous cannula MC2X, 18 Fr elongated-one-piece arterial cannula and 16 Fr DLP left atrial vent, Medtronic, Minn) and connected to an ECMO circuit (centrifugal pump, nonpulsatile flow, Cardiohelp MECC set, Quadrox oxygenator, Maquet, Rastatt, Germany). The VA-ECMO circuit had a shunt between the arterial and venous tubing. Clamping the inlet and outlet tubing while opening the shunt enabled rapid pressure and volume equilibration (22,27). Flows in the pulmonary artery and the ECMO circuit were measured with appropriate transit time ultrasonic flow probes (PAU and ME9 PXL Tubing flow sensors, respectively; Transonic, Ithaca, NY) and were monitored in real time to assist in volume and pump speed management (see later). Ventricular epicardial electrodes (MYO/Wire Temporary Atrial Cardiac Pacing Wires, A&E Medical Corporation, Farmingdale, NJ), and passive pleural drains were placed. The pericardium, sternum, and wound layers were closed. Intermittent heparin boluses were used to keep an activated clotting time >180 s. During ECMO, tidal ventilation was continued with a respiratory rate fixed at 16/min and fraction of inspired oxygen of 0.21. The sweep gas flow (100% O2) was adjusted to keep arterial Po2 and Pco2 in the normal range (ABL90Flex, Radiometer Medical ApS, Brønshøj, Denmark).

Pressure measurement and data acquisition

Intravascular and airway pressures were measured using transducers (xtrans, Codan Medical, Lensahn, Germany) and a multimodular monitor (S/5 Critical Care Monitor, Datex-Ohmeda, GE Healthcare, Helsinki, Finland) which also provided continuous electrocardiography and end-tidal Pco2. Output from the monitor and flow probes was recorded at 100 Hz in a data acquisition system (LabVIEW, National Instruments, Austin, Tex), and processed offline using customized analysis software (Soleasy, Alea Solutions, Zürich, Switzerland). The tip of the catheter used for right atrial pressure measurement, the venous drainage cannula, the inlet port of the ECMO pump, and all pressure transducers were fixed to the height of the mid-RA and verified by open chest palpation. Pressures were zeroed against the atmosphere and two-point calibrated using a water manometer. Flow probes were zeroed and calibrated electronically. Baseline drift for pressure and flows was checked at the end of the experiment.

Fluid administration, volume state, and ECMO pump speed

During surgery, Ringer's lactate was infused at a rate of 10 mL/kg/h and thereafter reduced to 2 mL/kg/h. Hydroxyethyl starch (HES, 6% Voluven, Fresenius Kabi, Bad Homburg, Germany) was supplemented for measured blood loss during surgery (150 ± 109 mL). After closing the chest and allowing a stabilization period of 30 min ECMO flow (QECMO) was adjusted to achieve a mixed venous O2 saturation (Svo2, measured in the RA) of 50% (lower normal limit for pigs). During this period, if necessary, HES was added in 50 mL boluses to allow sufficient QECMO to reach the Svo2 target, and to avoid RA collapse during tidal ventilation (total volume of HES, including that for blood loss replacement was 197 ± 199 mL). After this stage, defined as Euvolemia, no more HES was allowed.

Experimental protocol

Ventricular fibrillation was induced by high rate pacing (1000 bpm, ventricular electrical output 18 mV, Pace 203, Osypka, Berlin, Germany). The protocol consisted of eight experimental conditions: Euvolemia was followed by three conditions of stepwise increasing rates of norepinephrine infusion (0.05, 0.125, and 0.2 μg/kg/min, each beginning with a bolus of 5 μg/kg [Vasoconstriction 1-3, respectively]), with study measurements starting after 5 min at each infusion rate. After completing measurements at Vasoconstriction 3, the norepinephrine rate was halved and 3 mins later discontinued completely, entering a state of Post Vasoconstriction. This was followed by three conditions of stepwise Volume Expansion (VE1-3) where 10 mL/kg of Ringer's lactate was infused over three mins at each step, with study measurements starting after five mins (Fig. 1). After completing the measurements, the animals were killed in deep anesthesia by withdrawing the ECMO support.

Fig. 1:
Experimental protocol.

ECMO pump speed maneuvers and venous return curves

For each experimental condition, during tidal ventilation, Maintenance ECMO pump speed was adjusted to achieve a QECMO resulting in Svo2 of 50%. To find the Maximum QECMO achievable without provoking clinically apparent RA collapse, the pump speed was increased during expiratory hold while observing the real-time flow, displayed on screen, and the ECMO tubing for signs of fluttering. The Maintenance and Maximum ECMO pump speeds and 80%, 60%, and 50% thereof were applied during expiratory hold. Maneuvers lasted for 30 s, after which pump speed was reset to Maintenance and tidal ventilation for at least 1 min until blood pressure returned to baseline. Data were extracted (as mean over 2 s) 9 s into the airway pressure hold, after flow had reached its new steady state (4). To properly characterize the vascular return function, knowing from a previous experiment that unapparent closing conditions may occur (22), venous return curves of RAP-QECMO data pairs were constructed after excluding all maneuvers displaying vascular collapse in the offline analysis (independently assessed by authors PWM, AH, DB). Maximum achievable QECMO, however, was analysed with closing conditions included. To quantify the effect of the left atrial vent, in six animals in Euvolemia, Vasoconstriction 3, and VE3, maneuvers with maximum and 50% pump speed were repeated with the vent closed. To quantify the shift of the venous return curves between Vasoconstriction 3 and VE3, QECMO was calculated for standard RAP, representing the mean of all conditions. Oxygen delivery (at Maintenance and Maximum QECMO) and oxygen consumption (VO2; at Maintenance QECMO) were calculated using standard formulas for arterial and mixed venous blood oxygen content.

Determination of MSFP

MSFP was determined after the pump speed maneuvers in each condition in a Stop flow maneuver(22). The ECMO circuit was clamped with open shunt in expiratory hold (22). Flow was resumed after 30 s or if signs of a reflex-mediated increase in arterial blood pressure (ABP) were seen. MSFP was taken as the mean value of RAP during 2 s of equilibrium defined from ABP nadir (22). At least 3 min were allowed for blood pressure to return to baseline. Stability of MSFP was studied at Vasoconstriction 3 and VE3 by repeating the MSFP determinations three times over 40 min.

Blood volume determination

Plasma volume was measured using indocyanine green dye dilution at Euvolemia, Vasoconstriction 3, and VE3, as previously described (4). Changes in plasma volume were calculated based on hematocrit and hemoglobin concentrations using Beaumont's method (16) when no direct plasma volume measurements were available.

Determination of vascular elastance, stressed, and unstressed volumes

Vascular elastance (Evasc) was determined at Vasoconstriction 3 and VE3 using the difference between MSFP obtained before and immediately after rapid bleeding of 9 mL/kg from the arterial ECMO tubing into a transfusion bag. The bled volume was retransfused. Systemic vascular elastance was calculated as Evasc = ΔMSFP/ΔBV. Stressed and unstressed volumes were determined from the x-intercept of the Evasc function (4,23).


Based on previous data, a sample size of eight animals was needed to detect a clinically relevant difference in MSFP of 1 mmHg (4). Data were analysed using SPSS software (Version 21; SPSS Inc., Chicago Ill). Two-way repeated measurements analysis of variance (ANOVA), within-subject factors treatment (vasoconstriction vs. volume expansion) and level of treatment intensity (0–3), was used to assess the effects of interventions. In case of significant interaction (treatment × intensity), each treatment was tested separately with one-way repeated measurements ANOVA to assess where changes occurred. Bonferroni correction was applied as appropriate. The vent effect was assessed with two-way repeated measurements ANOVA (within-subject factors condition [Euvolemia, Vasoconstriction 3 and VE 3] and pump speed [maximum vs. 50%]). Blood volumes, elastances and hemoglobin concentrations (Vasoconstriction 3 vs. VE3), and urine output during Vasoconstriction 1-3 vs. VE 1-3 were compared with paired t test. Generalized estimating equations [(GEE) first order auto-regressive working correlation matrix] was used to characterize the linear relations between flow versus pump speed, pressure head versus flow, MSFP versus time, and the venous return function flow versus RAP. Proportion of variance for these variables in individual animals was assessed as Pearson correlation coefficient squared (r2). Assumptions of equal variance and normality were assessed as studentized residuals < ± 3, visually by Q-Q plots and histograms, and by Kolmogorov–Smirnov testing.


Ventricular fibrillation could be achieved in all nine animals with complete cessation of pulmonary artery blood flow (7 ± 12 mL/min over all conditions). Opening or closing the vent did not affect QECMO (range of relative changes, open to closed, 98.4 ± 3.3% to 100.1 ± 0.7%), with no difference seen between conditions or pump speeds (P = 0.502 and 0.598, respectively). Pump function (mL/revolution) was highly linear over the experimental conditions and pump speeds used [r2 (median, range) for individual animals over Euvolemia, Vasoconstriction 3 and VE 3 0.999 (0.811 – 1.0), Supplemental Digital Content, Table e1 and Figure e1,].

Both treatments progressively increased MSFP. For the doses used, the effect was more pronounced for Volume Expansion. Mean arterial pressure was higher with Vasoconstriction. Hemoconcentration and hemodilution were seen with Vasoconstriction and Volume Expansion, respectively. The blood lactate increased together with VO2 despite maintenance of the target Svo2 (Table 1). Factors defining venous return were not different between Euvolemia and Post Vasoconstriction (Supplemental Digital Content, Table e2, Urine output was 2.6 ± 1.1 vs. 2.7 ± 1.4 mL/kg/h during Vasoconstriction 1-3 and Volume Expansion 1-3, respectively (n = 6, P = 0.832).

Table 1:
Hemodynamics at maintenance pump speed

Maximum ECMO flow

Both treatments increased maximum achievable QECMO. For the doses used, the effect was more pronounced for Volume Expansion, but this did not translate into higher DO2 compared to Vasoconstriction, due to concomitant hemodilution (Table 2).

Table 2:
Maximum flow

Venous return function

Signs of vascular collapse were observed in 17% of pump speed maneuvers with equal distribution between treatments. When pump speeds were varied, there was a linear negative correlation between QECMO and RAP [median for individual QECMO/RAP responses r2 0.975 (0.626–1.000); Table 3, Fig. 2]. MSFP and flow increased significantly with both treatments, and the respective VR curves were shifted to the right (Tables 1–2, Fig. 2; for VR plots with all data pairs included, see Supplemental Digital Content, Table e3, Figure e2, Increase in flow was less pronounced in Vasoconstriction compared to Volume Expansion (Table 2). VRdP was not different between conditions and at maintenance speed (Table 1 and 4). The response of resistance to venous return from Vasoconstriction was highly variable, with equal distribution of increasing, decreasing or unchanged RVR in individual animals (P = 0.445). Volume Expansion consistently and progressively reduced RVR (Tables 3–4, Fig. 2). The flow corresponding to the standard RAP of 2.8 mmHg was 2978 ± 1046 mL/min in Euvolemia and increased to 3529 ± 648 mL/min during Vasoconstriction 3 and to 6195 ± 1787 mL/min during VE3 (p < 0.0005, for details see Supplemental Digital Content, Table e4,

Table 3:
Venous return function
Fig. 2:
Venous return curves after exclusion of closing conditions.
Table 4:
Venous return function, closing conditions excluded

Pressure head versus QECMO

The relationship between pressure head (mean arterial pressure MAP minus RAP) and QECMO, was highly linear in all conditions. (r2 for individual animals (median, range) in Euvolemia: 0.983 (0.936–0.999); Vasoconstriction 3: 0.990 (0.969–1.000); and VE3: 0.965 (0.643–0.995). The resistance needed to be overcome by the pump was lower in VE3 as compared to Euvolemia and/or Vasoconstriction 3 (GEE, Table 5).

Table 5:
Pressure head versus QECMO per condition

Vascular elastance, stressed, and unstressed blood volumes

Compared to Euvolemia, Vasoconstriction increased MSFP and decreased total blood volume due to loss of plasma (Table 1 and 6, Fig. 3). Volume Expansion restored and increased the blood volume slightly above the base level at Euvolemia due to plasma expansion. Vasoconstriction resulted in higher vascular elastance than Volume Expansion. Vasoconstriction led to a leftward shift of the elastance curve, and unstressed volume was recruited into stressed volume. Volume Expansion shifted the elastance curve back to the right, through increases in both stressed and unstressed volumes (Table 6, Fig. 3).

Table 6:
Blood volumes and volume shifts
Fig. 3:
Vascular elastance derived from bleeding maneuvers at Vasoconstriction 3 and VE3 (Table 6).

Stability of effects

At Vasoconstriction 3 and VE3, repeated measurements of MSFP over 40 min showed a decline over time (mean 1.7 mmHg), with no difference between treatments (GEE, Supplemental Digital Content, Table e5, Changes in plasma volume under these conditions were small (0.3 ± 6.5% for Vasoconstriction 3, −2.5 ± 7.7% for VE3, P = 0.24).


We have applied the principles of venous return (13–15), verified in a series of experiments with (22) and without mechanical circulatory assist (4,27,31), to modern VA-ECMO treatment. We found that both vasoconstriction with norepinephrine and volume expansion increased MSFP and the maximum achievable QECMO with similar oxygen delivery. The effect of volume expansion on blood flow was larger than that of vasoconstriction. In our model, the ECMO pump replaces the cardiac function. The pump function was constant, as indicated by the linear relationship between pump flow and revolutions per minute (rpm). Accordingly, our results can be interpreted solely as changes in the circuit properties, as we have previously demonstrated (22). The evaluation of effects and mechanisms of vasoconstriction and volume expansion on ECMO flow is highly relevant for the clinical application of modern ECMO treatment.

The maximum ECMO flows in each condition were associated with imminent vascular collapse, which could not be observed clinically. The vascular collapse, when present, dissociated the QECMO-RAP relationship because RAP no more served as the backpressure for VR (22,34). Closing conditions via vascular waterfalls were recognized as the main limitation to further flow increase in the early seminal studies by Guyton (14,15). The venous return plots showed a strictly linear RAP-QECMO relationship (Fig. 2), as predicted by Guyton's model. As we had hypothesized, both Vasoconstriction and Volume Expansion allowed for increasing maximum ECMO flow and increased MSFP. In this sequential treatment study design, we observed larger effects regarding blood flow from volume expansion than from vasoconstriction. Both treatments thereby shifted the VR curve to the right. In addition, Volume Expansion also decreased resistance to venous return. As compared to Euvolemia, the increase in flow under Vasoconstriction was accompanied by decreased total blood volume (a leftward shift of the vascular elastance curve) and increased MSFP by recruitment of stressed volume from unstressed vascular volume (Fig. 3). Recruitment of stressed volume was modified by two phenomena. First, vasoconstriction with norepinephrine increased the vascular elastance (29) and thereby increased MSFP for the given stressed volume. We did not measure elastance in Euvolemia and therefore cannot quantify the elastance increase under Vasoconstriction. The value of elastance reported here under norepinephrine is larger than we found in a similar experiment under euvolemic conditions (4), and increasing elastances have been shown for different vasoconstrictors (1,11). Second, roughly a fifth of the plasma volume and therefore part of the recruited volume may have been lost by plasma leakage. Plasma leakage may be related to inflammation induced by use of extracorporeal circulation (20). Volume Expansion shifted the elastance curve further to the right. Stressed and unstressed volumes were both expanded, but to a lesser extent than expected from the large amount of volume infused (30 mL/kg). As urinary output remained stable between conditions, this rightward shift may have been limited by ongoing plasma leakage and/or increases in vessel bed diameter, as described below.

For both Vasoconstriction and Volume Expansion, the maximum flow increased. Maximum flows were often associated with vascular collapse. Under closing conditions, the MSFP-RAP pressure gradient does not reflect the driving pressure for venous return. When venous return was evaluated in conditions without signs of collapse, a decrease in RVR after volume expansion was evident, and explains the further increase in VR despite unchanged VRdP. Recruitment of stressed from unstressed volume via vasoconstriction of veins and venules can occur without changes in the resistance to venous return (9,10), which is here reproduced with an unchanged slope on average (representing RVR) of the Vasoconstriction VR curves (Tables 3–4, Fig. 2). Notably, individual responses to norepinephrine varied, exhibiting unchanged, rising of falling resistances. Such variable reactions of venous return to vasoconstriction have been reported earlier (18) and are clinically important, when VA-ECMO is used as support for severe heart failure, as increases in resistance and afterload may have detrimental effects (25). Maximum flow may have been influenced by cannula tip-vessel wall interaction (34) via centralization of blood volume from vasoconstriction. We have therefore estimated the increases in QECMO at a standardized RAP, to exclude artifacts from dissociated QECMO and RAP, which confirmed the flow increases and curve shifts. Volume Expansion showed progressively and uniformly lower RVR and pressure heads, allowing higher flows at stable VRdP. Despite higher flow from volume expansion, oxygen delivery was limited due to concomitant hemodilution, and the resulting DO2 was similar in the two treatments. Besides resistance changes, an additional mechanism may be at play with volume expansion. The linear pump function illustrates that the flow generated per rpm will depend on the variables of the Hagen-Poiseuille equation (17), i.e., flow is directly proportional to the pressure gradient (or head) and the fourth power of vessel radius, but inversely proportional to viscosity and tubing length. The pressure heads and RVR at Volume Expansion may have been influenced by viscosity changes due to hemodilution or -concentration. Whether the decreasing resistances are a direct vasodilatory effect of Volume Expansion after vasopressor weaning, as is clinically often seen (30), or if ongoing SIRS and instability of the experimental preparation were the cause of vasodilation, cannot be determined with certainty.

What are the clinical consequences of our findings? Operating close to the maximum possible flow brings a high risk of clinically unapparent closing conditions. Volume depletion and high airway pressure (4,32) may increase the likelihood of vessel closure or directly reduce venous return via elevated right atrial pressure (22,27,31). Preferential drainage from the inferior caval vein (which in pigs has an intrathoracic part exposed to pleural pressure) with a three-stage cannula may have promoted vessel collapse and dissociated flow from right atrial pressure (34) as compared to a right angle cannula in a previous experiment (22). In the absence of clear evidence for optimal hemodynamic supportive measures, the clinician's choice between volume expansion and vasoconstriction should be guided by the disease process and the expected physiological limits and effects of a treatment.

Vasoconstriction may allow increase in flow by recruitment of stressed volume and thereby decreasing the need to infuse volume, where the amount is associated with worsening outcome for patients on VA-ECMO (28). Such volume sparing effects may be of special value in cases of severe respiratory failure. The physiological reaction to vasoconstriction is much more variable and therefore less predictable than that of volume expansion. Especially the increases in resistance may have negative effects in patients with failing hearts supported with temporary mechanical assist. Here, prudent volume expansion to facilitate vasodilation may be appropriate. As the components of venous return are not easily measured, monitoring the true effects of vasoconstriction is demanding. As a compromise, ECMO blood flow may be kept as low as clinically reasonable and adverse effects of repeated vascular collapse need to be considered when high flows are necessary. Vasoconstriction and volume expansion, as used in this study, are equal regarding oxygen delivery. Both show a similar decline in MSFP over time, probably due to ongoing plasma leakage. Plasma leakage under vasoconstriction has been described (12). The upper limit for recruitment of unstressed volume into stressed volume using vasoconstriction is reported as 10 to 18 mL/kg (19). Up to 3/4 of volume expansion with crystalloids will be lost into the interstitial space over time and eventually impair tissue perfusion in case of severe oedema. In pilot animals for this study, we used higher doses of norepinephrine, which led to more pronounced leakage and unstable preparation with inability to sustain the Svo2 target. The recruitable reserve and the ongoing leakage may be further influenced by inflammation associated with ECMO (20), and must be taken into clinical consideration.


Lack of randomization between Volume Expansion and Vasoconstriction is a limitation. We chose the sequential use of norepinephrine followed by volume expansion to minimize shifts in blood volume before norepinephrine. In clinical use of ECMO vasoconstriction and volume expansion are often used simultaneously or consecutively, and their effects are modified by deterioration of the underlying disease, ongoing plasma leakage and inflammation. Alternatively, volume could have been expanded and then removed to facilitate randomization. Transfusion and bleeding was not possible due to lack of pig blood for volume expansion. The ECMO system used did not have a volume reservoir or allow for ultrafiltration. Adding a CRRT device would have further increased the technical complexity of an already challenging setup. In addition, the effects of ultrafiltration on equilibration between interstitial and intravascular space and on vasoregulation would have interfered with restoration of baseline intravascular volume state after fluid removal. Our sequential approach was a pragmatic compromise. The main determinants of venous return did not differ between the two baseline conditions Euvolemia and Post Vasoconstriction. This suggests that the conclusion regarding the basic mechanisms of vasoconstriction and volume expansion still hold true at least during an early clinical course on VA-ECMO.

We did not encounter clinical instability during this experiment and similar previous experiments (4,22). Urinary output was stable during both Vasoconstriction and Volume Expansion. We attribute the rising lactate to an ongoing inflammatory reaction—a known phenomenon on ECMO (20). This is supported by the increasing oxygen consumption in the course of the experiment. We cannot exclude gut ischemia or liver dysfunction due to venous outflow obstruction but consider them unlikely due to the clinical stability. The continuously fibrillating heart may also have contributed, but not to a quantitatively relevant extent.

We tested whether vasoconstriction and volume expansion could increase maximum achievable ECMO flow and assessed the underlying mechanisms. Our study was not designed to evaluate a treatment benefit of one approach over the other. We show that both volume expansion and vasoconstriction, when used in moderate doses, increase maximum achievable ECMO flow, with similar effects on DO2. We conclude that ECMO flow is primarily dependent on the factors governing venous return - in our view a central finding for anyone in clinical care of ECMO patients. To find an optimal treatment regarding outcome, the basic mechanisms should be understood. To what degree our findings can be translated to diseases other than cardiac arrest, like septic shock or severe pulmonary failure, is still to be explored. Particularly, our results cannot be extrapolated to treatment of respiratory failure using veno-venous ECMO.

We omitted an elastance measurement at Euvolemia, which could have provided interesting information as we found changing elastances between Vasoconstriction and Volume Expansion. As there is little doubt about the linear behavior of the elastance curve in the physiological range (1,4,10,22,23), a one-step change of blood volume seems warranted. Values presented here are in agreement with previous experiments performed by us and others (4,24) and increasing elastance with vasoconstrictors is well described (1,11).

We used low doses of norepinephrine as titration of higher doses in the pilot series led to progressive instability. Still, higher doses may have shown clearer results. Similar results to ours were found with much higher doses in endotoxemic pig models (9).

Statistical approach

Linear regression has been the standard method of describing venous return (27). We could reproduce our earlier findings using standardized RAP and Generalized Estimating Eqs. (22), which allowed statistical interference from repeated measurements.

Validity of RAP and VRdP

Increasing pump speed shifts volume away from the RA, progressively lowering RAP and increasing VRdP (22)—which is the difference of intravascular pressures over a vascular segment. The limit of maximum flow, however, is determined by transmural pressure (26,34): at closing conditions, the vascular wall interacts with the cannula tip causing flow to drop with subsequent build-up of pressure until wall and cannula separate anew and flow is restored (staccato flow) (34). This phenomenon is associated with a net increase in resistance. A RAP valid for calculation of VRdP can only be measured at the orifice of unobstructed flow in a multistage cannula (14,34). We therefore verified our main results by estimation of increasing flows at standard RAP, independently of VRdP.

Independently of these limitations, the corollary of our experiment is that in a circulation completely dependent on a mechanical pump, maximum pump output is determined by vascular factors. This should be taken into consideration in the clinical management.


In a circulation completely dependent on ECMO support, blood flow is directly dependent on the vascular factors that govern venous return—i.e., closing conditions, stressed vascular volume and the elastance and resistive properties of the vasculature.


Olgica Beslac, Kay Nettelbeck, Sandra Nansoz, and Michael Lensch deserve our gratitude for their expert technical assistance during the experiments and Hansjörg Jenni for his valuable support with the ECMO equipment.


1. Appleton CP, Lee RW, Martin GV, Olajos M, Goldman S. Alpha 1- and alpha 2-adrenoceptor stimulation: changes in venous capacitance in intact dogs. Am J Physiol 250:H1071–H1078, 1986.
2. Bartlett R. Physiology of Extracorpreal Life Support. In: Extracorporeal Cardiopulmonary Support in Critical Care, 4th ed. edited by Annich GL, W. MacLaren, G. Wilson, J. Bartlett, R. Michigan: Extracorporeal Life Support Organisation.
3. Berger D, Moller PW, Takala J. Reply to “Letter to the editor: why persist in the fallacy that mean systemic pressure drives venous return?”. Am J Physiol Heart Circ Physiol 311:H1336–H1337, 2016.
4. Berger D, Moller PW, Weber A, Bloch A, Bloechlinger S, Haenggi M, Sondergaard S, Jakob SM, Magder S, Takala J. Effect of PEEP, blood volume, and inspiratory hold maneuvers on venous return. Am J Physiol Heart Circ Physiol 311:H794–H806, 2016.
5. Caldini P, Permutt S, Waddell JA, Riley RL. Effect of epinephrine on pressure, flow, and volume relationships in the systemic circulation of dogs. Circ Res 34:606–623, 1974.
6. Cheng Y, Pan T, Ge M, Chen T, Ye J, Lu L, Chen C, Zong Q, Ding Y, Wang D. Evaluation of vasopressin for vasoplegic shock in patients with preoperative left ventricular dysfunction after cardiac surgery: a propensity-score analysis. Shock 50:519–524, 2018.
7. Chouihed T, Kimmoun A, Lauvray A, Laithier F-X, Jaeger D, Lemoine S, Maureira JP, Nace L, Duarte K, Albizzati S, et al. Improving patient selection for refractory out of hospital cardiac arrest treated with extracorporeal life support. Shock 49:24–28, 2018.
8. Combes A, Brodie D, Chen Y-S, Fan E, Henriques JPS, Hodgson C, Lepper PM, Leprince P, Maekawa K, Muller T, et al. The ICM research agenda on extracorporeal life support. Intensive Care Med 43:1306–1318, 2017.
9. Datta P, Magder S. Hemodynamic response to norepinephrine with and without inhibition of nitric oxide synthase in porcine endotoxemia. Am J Respir Crit Care Med 160:1987–1993, 1999.
10. Deschamps A, Magder S. Baroreflex control of regional capacitance and blood flow distribution with or without alpha-adrenergic blockade. Am J Physiol 263:H1755–H1763, 1992.
11. Drees JA, Rothe CF. Reflex venoconstriction and capacity vessel pressure-volume relationships in dogs. Circ Res 34:360–373, 1974.
12. Dubniks M, Persson J, Grande PO. Effect of blood pressure on plasma volume loss in the rat under increased permeability. Intensive Care Med 33:2192–2198, 2007.
13. Guyton AC. Determination of cardiac output by equating venous return curves with cardiac response curves. Physiol Rev 35:123–129, 1955.
14. Guyton AC, Lindsey AW, Abernathy B, Richardson T. Venous return at various right atrial pressures and the normal venous return curve. Am J Physiol 189:609–615, 1957.
15. Guyton AC, Lindsey AW, Kaufmann BN. Effect of mean circulatory filling pressure and other peripheral circulatory factors on cardiac output. Am J Physiol 180:463–468, 1955.
16. Harrison MH. Effects on thermal stress and exercise on blood volume in humans. Physiol Rev 65:149–209, 1985.
17. Kohler K, Valchanov K, Nias G, Vuylsteke A. ECMO cannula review. Perfusion 28:114–124, 2013.
18. Maas JJ, Pinsky MR, de Wilde RB, de Jonge E, Jansen JR. Cardiac output response to norepinephrine in postoperative cardiac surgery patients: interpretation with venous return and cardiac function curves. Crit Care Med 41:143–150, 2013.
19. Magder S. Vincent JL. How does volume make the blood go around. Springer, Annual Update in Intensive Care and Emergency Medicine 2015. Cham, Switzerland: 2015.
20. Millar JE, Fanning JP, McDonald CI, McAuley DF, Fraser JF. The inflammatory response to extracorporeal membrane oxygenation (ECMO): a review of the pathophysiology. Crit Care 20:387, 2016.
21. Mitzner W, Goldberg H. Effects of epinephrine on resisitive and compliant properties of the canine vasculature. J Appl Physiol 39:272–280, 1975.
22. Moller PW, Winkler B, Hurni S, Heinisch PP, Bloch A, Sondergaard S, Jakob SM, Takala J, Berger D. Right atrial pressure and venous return during cardiopulmonary bypass. Am J Physiol Heart Circ Physiol 313:H408–H420, 2017.
23. Nanas S, Magder S. Adaptations of the peripheral circulation to PEEP. Am Rev Respir Dis 146:688–693, 1992.
24. Ogilvie RI, Zborowska-Sluis D, Tenaschuk B. Measurement of mean circulatory filling pressure and vascular compliance in domestic pigs. Am J Physiol 258:H1925–H1932, 1990.
25. Ostadal P, Mlcek M, Kruger A, Hala P, Lacko S, Mates M, Vondrakova D, Svoboda T, Hrachovina M, Janotka M, et al. Increasing venoarterial extracorporeal membrane oxygenation flow negatively affects left ventricular performance in a porcine model of cardiogenic shock. J Transl Med 13:266, 2015.
26. Permutt S, Riley RL. Hemodynamics of collapsible vessels with tone: the vascular waterfall. J Appl Physiol 18:924–932, 1963.
27. Pinsky MR. Instantaneous venous return curves in an intact canine preparation. J Appl Physiol Respir Environ Exerc Physiol 56:765–771, 1984.
28. Schmidt M, Bailey M, Kelly J, Hodgson C, Cooper DJ, Scheinkestel C, Pellegrino V, Bellomo R, Pilcher D. Impact of fluid balance on outcome of adult patients treated with extracorporeal membrane oxygenation. Intensive Care Med 40:1256–1266, 2014.
29. Trippodo NC. Total circulatory capacity in the rat. Effects of epinephrine and vasopressin on compliance and unstressed volume. Circ Res 49:923–931, 1981.
30. Vane LA, Prough DS, Kinsky MA, Williams CA, Grady JJ, Kramer GC. Effects of different catecholamines on the dynamics of volume expansion of crystalloid infusion. Anesthesiology 101:1136–1144, 2004.
31. Versprille A, Jansen JR. Mean systemic filling pressure as a characteristic pressure for venous return. Pflugers Arch 405:226–233, 1985.
32. Vieillard-Baron A, Augarde R, Prin S, Page B, Beauchet A, Jardin F. Influence of superior vena caval zone condition on cyclic changes in right ventricular outflow during respiratory support. Anesthesiology 95:1083–1088, 2001.
33. Wengenmayer T, Rombach S, Ramshorn F, Biever P, Bode C, Duerschmied D, Staudacher DL. Influence of low-flow time on survival after extracorporeal cardiopulmonary resuscitation (eCPR). Crit Care 21:157, 2017.
34. Wenger RK, Bavaria JE, Ratcliffe MB, Bogen D, Edmunds LH Jr. Flow dynamics of peripheral venous catheters during extracorporeal membrane oxygenation with a centrifugal pump. J Thorac Cardiovasc Surg 96:478–484, 1988.

Crystalloid solutions; extracorporeal membrane oxygenation; hemodynamics; mean systemic filling pressure; norepinephrine; stressed volume; vasoconstrictor agents; venous return; ABP; arterial blood pressure; ANOVA; analysis of variance; DO2; delivery of oxygen; ECMO; extracorporeal membrane oxygenation; GEE; generalized estimating equation; HES; hydroxyethyl starch; MAP; mean arterial pressure; MSFP; mean systemic filling pressure; PEEP; positive end-expiratory pressure; RAP; right atrial pressure; rpm; revolutions per minute; Svo2; mixed venous oxygen saturation; VE; volume expansion; VO2; oxygen consumption; VR; venous return; VRdP; Venous return driving pressure (=MSFP – RAP)

Supplemental Digital Content

Copyright © 2018 by the Shock Society