Functional evaluation of sublingual microcirculation indicates successful weaning from VA-ECMO in cardiogenic shock
Veno-arterial extracorporeal membrane oxygenation (VA-ECMO) is increasingly adopted for the treatment of cardiogenic shock (CS). However, a marker of successful weaning remains largely unknown. Our hypothesis was that successful weaning is associated with sustained microcirculatory function during ECMO flow reduction. Therefore, we sought to test the usefulness of microcirculatory imaging in the same sublingual spot, using incident dark field (IDF) imaging in assessing successful weaning from VA-ECMO and compare IDF imaging with echocardiographic parameters.
Weaning was performed by decreasing the VA-ECMO flow to 50% (F50) from the baseline. The endpoint of the study was successful VA-ECMO explantation within 48 hours after weaning. The response of sublingual microcirculation to a weaning attempt (WA) was evaluated. Microcirculation was measured in one sublingual area (single spot (ss)) using CytoCam IDF imaging during WA. Total vessel density (TVDss) and perfused vessel density (PVDss) of the sublingual area were evaluated before and during 50% flow reduction (TVDssF50, PVDssF50) after a WA and compared to conventional echocardiographic parameters as indicators of the success or failure of the WA.
Patients (n = 13) aged 49 ± 18 years, who received VA-ECMO for the treatment of refractory CS due to pulmonary embolism (n = 5), post cardiotomy (n = 3), acute coronary syndrome (n = 2), myocarditis (n = 2) and drug intoxication (n = 1), were included. TVDssF50 (21.9 vs 12.9 mm/mm2, p = 0.001), PVDssF50 (19.7 vs 12.4 mm/mm2, p = 0.01) and aortic velocity–time integral (VTI) at 50% flow reduction (VTIF50) were higher in patients successfully weaned vs not successfully weaned. The area under the curve (AUC) was 0.99 vs 0.93 vs 0.85 for TVDssF50 (small vessels) >12.2 mm/mm2, left ventricular ejection fraction (LVEF) >15% and aortic VTI >11 cm. Likewise, the AUC was 0.91 vs 0.93 vs 0.85 for the PVDssF50 (all vessels) >14.8 mm/mm2, LVEF >15% and aortic VTI >11 cm.
This study identified sublingual microcirculation as a novel potential marker for identifying successful weaning from VA-ECMO. Sustained values of TVDssF50 and PVDssF50 were found to be specific and sensitive indicators of successful weaning from VA-ECMO as compared to echocardiographic parameters.
KeywordsCardiogenic shock VA-ECMO Microcirculation Incident dark field imaging Sublingual CytoCam Weaning Cardiac recovery
Area under the curve
Decreasing the veno-arterial extracorporeal membrane oxygenation flow to 50% of the baseline flow
Intensive care unit
Incident dark field
Left ventricular ejection fraction
Not successfully weaned
Proportion of perfused vessels
Perfused vessel density
Single-spot measurements of perfused vessel density
Single-spot measurements of perfused vessel density during 50% flow reduction
Receiver operating characteristic
Sidestream dark field
Tricuspid annular planer systolic excursion
Tissue Doppler lateral mitral annulus peak systolic velocity
Total vessel density
Single-spot measurements of total vessel density
Single-spot measurements of total vessel density during 50% flow reduction
Veno-arterial extracorporeal membrane oxygenation
Aortic velocity–time integral at 50% flow reduction
Cardiogenic shock (CS) associated with cardiac pump failure results in a state of inadequate tissue perfusion, which leads to organ failure with a mortality rate between 50 and 80% . Veno-arterial extracorporeal membrane oxygenation (VA-ECMO) is considered a lifesaving treatment that is increasingly used for the treatment of critically ill patients that have experienced CS [2, 3, 4]. However, mortality rates remain high, with reports of up to 44% mortality despite the use of VA-ECMO [1, 5].
Current strategies for weaning from VA-ECMO are ongoing, largely unknown and based on empirical evidence [6, 7, 8]. Most of the current markers of weaning from VA-ECMO are based on echocardiography, such as aortic velocity–time integral (VTI), left ventricular ejection fraction (LVEF), and tissue Doppler lateral mitral annulus peak systolic velocity (TDSa) . However, performing high-quality echocardiography in critically ill patients requires specialized training and is relatively costly .
In a recent study, we found that the initial inability of VA-ECMO to recruit the microcirculatory alteration associated with CS predicts adverse outcomes following VA-ECMO treatment irrespective of improved systemic hemodynamic parameters . This is based on the concept that the success of resuscitation from states of circulatory shock is the normalization of microcirculatory and tissue perfusion [12, 13].
It is known that there is possibly a loss of coherence between the systemic and microcirculation, which can occur in states of shock and resuscitation . Previous studies measuring sublingual microcirculation using hand-held video microscopy have shown impairment of sublingual microcirculation to be associated with CS [12, 13, 15]. In addition, studies have found that sustained microcirculatory perfusion by VA-ECMO as detected by handheld video microscopes is associated with lower morbidity and even mortality [12, 14].
Our hypothesis was that successful weaning is associated with sustained microcirculatory function during ECMO flow reduction. Therefore, we sought to test the usefulness of microcirculatory imaging in the same sublingual spot, using incident dark field (IDF) imaging in assessing successful weaning from VA-ECMO and to compare IDF imaging with echocardiographic parameters .
Transthoracic echocardiography (TTE) and/or transesophageal (TEE) echocardiography was performed during weaning attempts using the CX50 ultrasound system (Philips Medical System, Best, The Netherlands). Pulsed-wave and continuous-wave Doppler signals were recorded at a sweep speed of 50–100 mm/s. Color Doppler recordings were optimized for display with the color velocity scale at ± 60 (50–70 cm/s) during the entire study.
All echocardiograms were analyzed by two experienced echo cardiologists (OS and SA), in accordance with published guidelines , using the QLAB quantification software (Philips Healthcare, Best, The Netherlands). Aortic VTI was measured by manually tracing the spectral envelope of continuous-wave Doppler in the apical 5-chamber or 3-chamber view. The LVEF was visually estimated from apical views. The right ventricular function was assessed by measuring the tricuspid annular planer systolic excursion (TAPSE) from the M-mode images in the apical 4-chamber view. Tissue Doppler lateral mitral annulus peak systolic velocity (TDSa) was also measured when feasible.
Weaning from VA-ECMO was initiated in patients with persistently stable hemodynamics (mean arterial pressure > 60 mmHg, lactate < 2 mmol/L and mixed venous saturation values > 65%) and with persisting arterial pulsatility wave on the monitor under low doses of inotropic support (Fig. 1). Weaning was performed by lowering the blood flow to 50% of the baseline value under hemodynamic and echocardiographic surveillance. Persisting hemodynamic stability was defined as aortic VTI > 10 cm and estimated LVEF > 20–25% [7, 9]. Patients who recovered from severe cardiac dysfunction and who tolerated the weaning attempt were considered for device removal .
Sublingual microcirculation was measured independently of echocardiography data during the same weaning attempt, just before or after echocardiography, at baseline (100%) VA-ECMO flow (F100), and after reducing VA-ECMO flow to 50% of the baseline flow (F50) and at returning VA-ECMO flow to baseline after 2 minutes (F100). We performed these measurements just before or after the classical weaning attempt using echocardiography. These microcirculatory data were not used to drive ECMO management.
Such a weaning attempt was followed by microcirculatory measurements during a weaning attempt, which took a maximum of 10 minutes in total. All aspects of microcirculatory measurements were performed as stable video recordings with a duration of 3–5 seconds by placing the CytoCam IDF imaging camera (Braedius Medical, Huizen, The Netherlands)  in the same sublingual area during the entire procedure, with total vessel density measured at a single spot (TVDSS).
The IDF device consists of a computer-controlled, high-resolution image sensor in combination with a specifically designed microscope lens at the end of an image guide, covered by a disposable sterile cap. Placing the tip of the guide to the sublingual tissue surface provides high-resolution images of the microcirculation where red blood cells can be clearly visualized flowing through the microvessels.
This new iteration of the device with improved optics detects 30% more sublingual vessels than the previous generation microscope [17, 19, 20]. Microcirculatory parameters are quantified by analyzing the movies using specialized image processing software (Automated Vascular Analysis (AVA)) .
Two microcirculation experts (SA and GG) independently analyzed all microcirculation parameters based on international consensus on the quantification of sublingual microcirculatory alterations . The images were analyzed to determine the functional parameters of large microvessels (> 25 μm) and small vessels (≤ 25 μm). These parameters consisted of the TVD (mm/mm2); perfused vessel density (PVD (mm/mm2)); and proportion of perfused vessels (PPV (%)) in accordance with international consensus guidelines related to the quantification of such microcirculatory images . Microcirculatory measurements were compared with echocardiography parameters (LVEF, aortic VTI and TDSa, if available) and used to evaluate the last weaning procedure as described by Aissaoui et al. .
Categorical variables are presented as frequencies and percentages. Continuous variables are presented as the mean ± standard deviation (SD). Continuous variables were compared with the Mann-Whitney U test. For comparisons within the same group, the microcirculatory parameters of patients at different time points were analyzed using the Friedman and Wilcoxon test. To compare the microcirculatory parameters of patients successfully weaned (SW) and not successfully weaned (NSW) at different time points, the generalized linear model repeated measurements test was used. Spearman’s correlation analysis was used to compare the correlation between echocardiographic and microcirculatory parameters in SW and NSW patients. Statistical significance was defined by a p value <0.05. Analyses were performed using SPSS version 184.108.40.206 (SPSS, IBM, Armonk, NY, USA) and MedCalc (MedCalc, Ostend, Belgium) software.
Baseline characteristics of the microcirculation of 13 patients observed during weaning attempts
Successful weaning (n = 10)
Non-successful weaning (n = 3)
Number of patients in each group
56 ± 17
41 ± 16
49 ± 18
Days on ICU
Days of ECMO
1.3 ± 0.7
2.0 ± 1.7
1.5 ± 0.5
The global hemodynamics were not significantly different between SW and NSW patients during weaning attempts (Additional file 2: Figure S1A and S1B). Successful and unsuccessful weaning was classified according to echocardiographic assessment as described previously. The results from the microcirculation measurements showed that in SW patients, TVDss, PVDss and PPVss measured in the same sublingual area maintained their values prior to the weaning attempt, whereas these values decreased in the patients who were NSW (Additional file 3: Figure S2A through Additional file 4: Figure S2F). TVDss, PVDss and PPVss were statistically significantly reduced following a flow reduction (from F100 to F50) in patients who were not NSW (TVDss all vessels p = 0.001; PVDss all vessels p = 0.01; PPVss all vessels p = 0.04; TVD small vessels p = 0.001; PVDss small vessels p = 0.01; PPV small vessels p = 0.03).
TVD of all vessels and TVD of small vessels were statistically reduced in the NSW patients during 50% VA-ECMO flow compared to no change or even increased values in SW patients. Examples of the recorded moving images of the sublingual microcirculation of the two categories of patients can be found in Additional file 8.
A comparison of the microcirculatory parameters with echocardiographic parameters values according to the published Aissaoui criteria for weaning from VA-ECMO  showed good correlation, especially with LVEF (r = 0.6214 and p = 0.01) (Additional file 7: Table S3).
Receiver operating characteristic (ROC) curves showed the area under the curve (AUC) was 0.99 vs 0.93 vs 0.85 for the TVDssF50 (small vessels) >12.2 mm/mm2, LVEF > 15% and aortic VTI > 11 cm (Additional file 9: Figure S3). Likewise, the AUC was 0.91 vs 0.93 vs 0.85 for the PVDssF50 (all vessels) > 14.8 mm/mm2, LVEF > 15% and aortic VTI > 11 cm.
The main finding in this study is that sustained sublingual microcirculation during VA-ECMO flow reduction in a convenient cohort sample of patients supported with VA-ECMO during cardiogenic shock, can provide a marker for the success of weaning from VA-ECMO. Cardiogenic shock affects all organs and compromises central hemodynamic cardiovascular function and consequently tissue perfusion. Currently used strategies for weaning from VA-ECMO are largely based on echocardiographic parameters. However, performing echocardiography in the ICU is challenging . The results of the present study show that functional parameters of microcirculation, including TVDss and PVDss, reflect recovery from cardiogenic shock and predict successful weaning from VA-ECMO. Patients who were successfully weaned had significantly higher baseline TVDss and PVDss compared to those of patients who were not successfully weaned. Even though global hemodynamics were comparable between patients with and without successful weaning, microcirculatory parameters were significantly different. The occurrence of such disassociation between macro-circulation and microcirculation, referred to as a loss of hemodynamic coherence, has been described before in other conditions of cardiovascular compromise [11, 23, 24, 25, 26, 27, 28].
The microcirculatory approach presented in this study could provide an alternative approach for the rapid assessment of cardiac recovery from cardiogenic shock during weaning attempts and/or in addition to echocardiographic evaluation. This approach could also be useful for the echocardiographic assessment of the left ventricular systolic function, especially in ICU patients with poor echo windows. Cavarocchi et al.  used a 4-stage strategy to evaluate 50% VA-ECMO blood flow, volume challenge and inotropic challenge during at least 1 hour of the continual monitoring of heart rate, blood pressure, and right ventricle (RV) and left ventricle (LV) function under transesophageal echocardiography. Weaning was considered successful when both LV and RV functions tolerated volume challenge and demonstrated inotropic reserve. However, this strategy required intravenous sedation to tolerate transesophageal echocardiography and an increased physical load in non-intubated patients throughout the weaning attempt with need for continuous therapeutic anticoagulation. A different approach involves the use of biomarkers as indicators of the success of weaning; however, such markers appear very late and seem inconclusive for determining the success of weaning from VA-ECMO. The usefulness of biomarkers in weaning from VA-ECMO, therefore, remains controversial. In line with this limitation, Luyt et al.  reported that in patients with refractory cardiogenic shock receiving VA-ECMO support, early measurements of cardiac biomarkers are not useful for identifying those who would recover.
In a VA ECMO weaning study, Aissaoui et al. measured left ventricular functional parameters (e.g., LVEF, VTI, TDSa) and found these to be good predictors of successful weaning. However, such echocardiographic assessment is limited to the evaluation of the left heart function under the condition that there are sufficient windows for analysis . The echocardiography parameters used in these weaning attempts, however, do not provide information about the right heart function, systemic hemodynamics or tissue perfusion, which also deteriorate as a consequence of cardiogenic shock. Our study shows that during weaning attempts, recovery from cardiogenic shock is revealed in the microcirculation, which agrees with total cardiac recovery upon echocardiography.
Monitoring the microcirculation using direct vital imaging with handheld microscopy has the potential to be the technique of choice to assess tissue perfusion in different phases of shock [27, 30]. The study described in this paper illustrated that the assessment of sublingual microcirculation and the echocardiographic evaluation of cardiac function was acceptably matched in discriminating between patients who were and were not successfully weaned. Microcirculatory evaluation was rapid since alterations were observed within 2 minutes following a lowering of ECMO flow.
Physiologically, this fast-adapting mechanism can also be evaluated from fractional flow reserve (FFR) measurements in coronary angiography. In these measurements, the hyperemic phase after the resolution of stenosis also occurs within 2 minutes . Several studies have been performed on microcirculatory alterations during ECMO [32, 33] with differing results concerning the relationship between global hemodynamics to the microcirculation [23, 34, 35, 36, 37, 38]. In a recent study, however, we found a significant predictive value of sublingual measured perfused vessel density in VA-ECMO patients for survival in the ICU . However, to date, sublingual measurements have not been employed to guide weaning from VA-ECMO.
The authors acknowledge the following limitations. First, this was a single-center observational study with a small population of patients with various underlying diseases causing cardiogenic shock. Furthermore, analyses of echocardiography and global hemodynamics, together with microcirculatory parameters measured in the same sublingual area, were performed only in patients deemed eligible to wean. This meant that we could not perform microcirculatory measurements without echocardiography for weaning attempts in patients under VA-ECMO support because of the observational nature.
The methods of measurement and evaluation of the microcirculation remain sensitive to artifacts and technical limitations. The strength of our study from a methodological perspective, however, lies in the fact that all measurements were performed in the same area (single spot). This methodology allowed us to assess the response of single microvessels to changes in pump settings as against comparing the mean value of microcirculation parameters of images at different locations and at different time points.
We found that the functional microcirculatory parameters measured sublingually using IDF imaging (TVDssF50 and PVDssF50) during weaning attempts for patients from VA-ECMO showed essential alterations within 2 minutes and prediction of cardiac recovery after cardiogenic shock. Future clinical and possible crossover studies should be designed in larger study populations undergoing VA-ECMO for monitoring microcirculation to guide weaning attempts.
Veno-arterial extracorporeal membrane oxygenation (VA-ECMO) use is a last option for survival in many types of cardiogenic shock (CS).
Conventional weaning from VA-ECMO is guided by echocardiographic parameters such as the aortic velocity − time integral on continuous wave Doppler recordings from left ventricular outflow tract and by assessing improvement in left ventricular ejection fraction. Echocardiographic measurements are not easily obtained in the ICU settings. On the other hand, a novel imaging technique, dark field imaging of the microcirculation, is quite feasible in almost all patients in the ICU.
Identified sublingual microcirculation is a novel potential marker for identifying successful weaning from VA-ECMO. Sustained values of single-spot measurements of total vessel density during 50% flow reduction (TVDssF50) and single-spot measurements of perfused vessel density during 50% flow reduction (PVDssF50) were found to be specific and sensitive indicators of successful weaning from VA-ECMO as compared to echocardiographic parameters. Therefore, weaning from VA-ECMO could be performed by imaging of the microcirculation using simple markers of tissue perfusion during weaning attempts.
We would like to thank R.T. van Domburg, PhD (Department of Cardiology, Erasmus MC Rotterdam), W. Rietdijk, PhD (Department of Intensive Care, Erasmus MC Rotterdam) for their statistical advice, P.A. Cummins from the editorial service of the Thoraxcenter Rotterdam for his English editing and Yasin Ince (Office manager at Academic Medical Center of Amsterdam) for his creation of the same spot sublingual microcirculation images (Fig. 3).
Availability of data and materials
Data are available from the author on reasonable request.
SA contributed in writing this manuscript, carried out the analysis of the microcirculatory measurements at the bedside and drafted the manuscript. He analyzed the clips and echocardiograms independently. He did the statistical analysis under supervision of Dr R.T. van Domburg and prepared the tables and the figures. He also analyzed the data from global hemodynamics, included the patients into the study for subsequent measurements at the bedside and revised the manuscript. DDRM contributed in writing of this manuscript. He helped in preparing the tables and figures after statistical analysis was performed under supervision of Dr R.T. van Domburg. He also participated in styling and writing the “Discussion” section of the manuscript, interpretation of the results and revising this manuscript. KC participated in the design of the study and interpretations of the echocardiography data and participated in the writing and revision of the manuscript. RJvT participated in the design of the study, interpretation of the data and revision of the manuscript. OIS participated in the analysis of the echocardiography data and participated in the writing and revision of the manuscript. GG participated in the design of the study, contributed to interpretation of the data and revised the manuscript. AS participated in the design of the study, contributed to interpretation of the data and revised the manuscript. RJvT participated in the design of the study, interpretation of the data and revision of the manuscript. LSJ participated in coordination during inclusion of the patients in the ICU and revision of the manuscript. AL participated in analysis of the clips and microcirculation measurements at the bedside and helped to revise the manuscript. DG contributed to the conception of the study, participated in its design and coordination and helped to draft the manuscript. FZ contributed to the conception of the study, helped to draft the manuscript and revised the manuscript critically for important intellectual content. CI conceived the study design and contributed during inclusion, analyzing, discussion, interpretations of the microcirculatory data, writing and revising this manuscript. All authors read and approved the final manuscript.
Ethics approval and consent to participate
The protocol was approved by the Medical Ethical Committee of Erasmus Medical Center of Rotterdam, the Netherlands (NL45915.078.13).
Consent for publication
Written informed consent was obtained from either patients or patient representatives.
Dr. Ince has developed SDF imaging and is listed as inventor on related patents commercialized by Micro Vision Medical (MVM) under a license from the Academic Medical Center (AMC). He has been a consultant for MVM in the past but has not been involved with this company for more than 5 years now and does not hold shares. Braedius Medical, a company owned by a relative of Dr. Ince, has developed and designed a handheld microscope called CytoCam-IDF imaging. Dr. Ince has no financial relationship of any sort with Braedius Medical, i.e., has never owned shares or received consultancy or speaker fees from Braedius Medical. Dr. Reis Miranda received speaking fees from NovaLung. All other authors state that they have no conflict of interest.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
- 16.Lang RM, Badano LP, Mor-Avi V, Afilalo J, Armstrong A, Ernande L, Flachskampf FA, Foster E, Goldstein SA, Kuznetsova T, et al. Recommendations for cardiac chamber quantification by echocardiography in adults: an update from the American Society of Echocardiography and the European Association of Cardiovascular Imaging. J Am Soc Echocardiogr. 2015;28(1):1–39. e14.CrossRefPubMedGoogle Scholar
- 20.Gilbert-Kawai E, Coppel J, Bountziouka V, Ince C, Martin D, Caudwell Xtreme E, Xtreme Everest 2 Research G. A comparison of the quality of image acquisition between the incident dark field and sidestream dark field video-microscopes. BMC Med Imaging. 2016;16(1):10.CrossRefPubMedPubMedCentralGoogle Scholar
- 33.Petroni T, Harrois A, Amour J, Lebreton G, Brechot N, Tanaka S, Luyt CE, Trouillet JL, Chastre J, Leprince P, et al. Intra-aortic balloon pump effects on macrocirculation and microcirculation in cardiogenic shock patients supported by venoarterial extracorporeal membrane oxygenation. Crit Care Med. 2014;42(9):2075–82.CrossRefPubMedGoogle Scholar
- 34.Bienz M, Drullinsky D, Stevens LM, Bracco D, Noiseux N. Microcirculatory response during on-pump versus off-pump coronary artery bypass graft surgery. Perfusion. 2016;31(3):207-15.Google Scholar
- 38.Yuruk K, Bezemer R, Euser M, Milstein DM, de Geus HH, Scholten EW, de Mol BA, Ince C. The effects of conventional extracorporeal circulation versus miniaturized extracorporeal circulation on microcirculation during cardiopulmonary bypass-assisted coronary artery bypass graft surgery. Interact Cardiovasc Thorac Surg. 2012;15(3):364–70.CrossRefPubMedPubMedCentralGoogle Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.