Korean J Thorac Cardiovasc Surg 2012;45:236-241
□ Clinical Research □
ISSN: 2233-601X (Print) ISSN: 2093-6516 (Online)
− 236 −
Department of Thoracic and Cardiovascular Surgery, Seoul National University Bundang Hospital, Seoul National University College of Medicine
†This study was presented at the 43th Annual Meeting of the Korean Society for Thoracic and Cardiovascular Surgery, Gwangju, Korea.
Received: November 22, 2011, Revised: December 18, 2011, Accepted: January 25, 2012
Corresponding author: Cheong Lim, Department of Thoracic and Cardiovascular Surgery, Seoul National University Bundang Hospital, Seoul
National University College of Medicine, 82 Gumi-ro 173beon-gil, Bundang-gu, Seongnam 463-707, Korea
(Tel) 82-31-787-7134 (Fax) 82-31-787-4050 (E-mail) email@example.com
C The Korean Society for Thoracic and Cardiovascular Surgery. 2012. All right reserved.
CC This is an open access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creative-
commons.org/licenses/by-nc/3.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the
original work is properly cited.
Prognosis in the Patients with
Prolonged Extracorporeal Membrane Oxygenation
Tae-Hun Kim, M.D., Cheong Lim, M.D., Il Park, M.D.,
Dong Jin Kim, M.D., Yochun Jung, M.D., Kay-Hyun Park, M.D.
Background: Prolonged usage of extracorporeal membrane oxygenation (ECMO) may induce multi-organ failure.
This study is aimed to evaluate prognostic factors in the patients with ECMO. Also, the prognosis of ECMO with
Kidney Injury Network Scoring system is studied. Materials and Methods: From May 2005 to July 2011, 172 cas-
es of ECMO were performed. The cases of perioperative use of ECMO were excluded. Renal failure patient and
younger than 15 years old one were also excluded. As a result, 26 cases were enrolled in this study. Male pa-
tients were 15 (57.7%), and mean age was 56.57±17.03 years old. Demographic data, ECMO parameters, weaning
from ECMO, and application of continuous renal replacement therapy are collected and Acute Kidney Injury
Network (AKIN) scores were evaluated just before ECMO and day 1, day 2 during application of ECMO. Results:
Venoarterial ECMO was applied in 22 cases (84.6%). The reasons for applications of ECMO were cardiac origin in
21 (80.8%), acute respiratory distress syndrome in 4, and septic shock in 1 case. Successful weaning from ECMO
was achieved in 15 cases (57.7%), and survival discharge rate was 9 cases (34.6%). Mean duration of application
of ECMO was 111.39±54.06 hours. In univariate analysis, myocarditis was independent risk factors on weaning
failure. Using the receiver operating characteristic curve, level of hemoglobin on 24 hours after ECMO, and base
excess on 48 hours after ECMO were showed more than 0.7. AKIN score was not matched the prognosis of the
patients with ECMO. Conclusion: In our study, the prognosis of the patients with myocarditis was poor.
Hemoglobin level at first 24 hours, and degree of acidosis at 48 hours were useful methods in relating with prog-
nosis of ECMO. AKIN scoring system was not related with the prognosis of the patients. Further study for prog-
nosis and organ injury during application ECMO may be needed.
Key words: 1. Prognosis
2. Extracorporeal membrane oxygenation
According to the data from Extracorporeal Life Support
Organization registry, successful weaning rate from ex-
tracorporeal membrane oxygenation (ECMO) is 57%, and
overall survival discharge rate is 46% in adult patients (Fig.
1) . Data from Health Insurance Review and Assessment
Service of Korea shows total cases of ECMO have been in-
creasing since 2006 (Fig. 2). Prolonged operation of ECMO
may produce many complications. Among them, organ dam-
ages may play an important role for the prognosis of the pa-
tients with ECMO. New devices, reducing the possibility of
Prognosis in Prolonged ECMO
− 237 −
Table 1. Acute Kidney Injury Network scoring system
Stage Creatinine criteriaUrine output criteria
Increase in serum creatinine ≥0.3 mg/dL or increase ≥1.5 times from
Increase in serum creatinine ≥2-fold from baseline
Increase in serum creatinine ≥3-fold from baseline or serum creatinine
≥4.0 mg/dL and initation of renal replacement therapy
≤0.5 mL/kg/hr for more than 6 hr
≤0.5 mL/kg/hr for more than 12 hr
≤0.3 mL/kg/hr for more than 24 hr
Fig. 1. Overall outcomes in adult extracorporeal membrane oxy-
genation patients. Data from the Extracorporeal Life Support
Organization registry (till July 2011). ECLS, extracorporeal life sup-
ports; ECPR, extracorporeal cardiopulmonary resuscitation (Modified
from Dalton HJ. Respir Care 2011;56:1445-53, with permission
from Daedalus Enterprises Inc.) .
Fig. 2. Yearly cases of extracorporeal membrane oxygenation in
Korea. A data from Health Insurance Review and Assessment
Service of Korea.
complications, have been developed and used. But, there is
no perfect device that would not damage organs. Recently,
some authors reported that scoring system for renal injury is
good predictor for ECMO patients . We reviewed our cas-
es of ECMO to find the prognostic factor. To find the rela-
tionship between organ damages and prognosis of the patients
with ECMO, the known scoring system was applied, which
was proposed by Acute Kidney Injury Network group (Table
MATERIALS AND METHODS
From May 2005 to July 2011, 172 cases of ECMO were
performed in Seoul National University Bundang Hospital.
Total cases of ECMO that had operated more than 48 hours
are 76. Because of the possibility of cardiopulmonary bypass
effect during open heart surgery, the cases of perioperative
application of ECMO were excluded. Known renal failure pa-
tients with hemo-dialysis were also excluded. Severity of dis-
ease classification system (SAPS) II was designed to estimate
the severity of disease for patients , who admitted in-
tensive care unit aged 15 or more, cases of younger than this
were also excluded. Finally, twenty-six patients were enrolled
in this study (Fig. 3). Male patients were 15 (57.7%), and fe-
male 11 (42.3%). Mean age was 56.57 (standard deviation,
±17.03) (range, 20 to 84) years old. Patients were divided in-
to two groups, ECMO weaning success group and failure
group. Body weight, smoking history, presence of diabetes,
hypertension, and history of cerebrovascular accident were
checked. Hemoglobin and serum creatinine levels at the time
of admission, 24 and 48 hours after initiation of ECMO were
checked. Platelet counts, level of aspartate aminotransferase
(AST), total bilirubin, glucose, and albumin 24 and 48 hours
after initiation of ECMO were also checked. Base excess, re-
flecting degree of metabolic acidosis, mean arterial pressure,
Tae-Hun Kim, et al
− 238 −
Table 2. Comparison between successful weaning of extracorporeal
membrane oxygenation and weaning failure groups
Body weight (kg)
Fig. 3. Inclusion and exclusion criteria and distribution by sex.
ECMO, extracorporeal membrane oxygenation; CRF, chronic renal fail-
ure; HD, hemodialysis; SAPS, severity of disease classification system.
Table 3. The results of Acute Kidney Injury Network (AKIN) scor-
ing system in both groups
AKIN0-hours 0, 1/2, 3
AKIN24-hours 0, 1/2, 3
AKIN48-hours 0, 1/2, 3
flows of ECMO at the time of starting, 24 and 48 hours of
ECMO were investigated. Regarding to ECMO, changes of
body weight after 48 hours, the amounts of transfusion, types
(venoarterial or venovenous), insertion place, and relation
with cardiopulmonary resuscitation, operating duration, and
application of continuous renal replacement therapy were
evaluated. Vasoactive agents, sequential organ failure assess-
ment (SOFA), and SAPS II scores were calculated. Acute
Kidney Injury Network (AKIN) scores were calculated and
applied for each stage of initiation; AKIN0-hours, 24;
AKIN24-hours, and 48 hours of ECMO; AKIN48-hours (Table 1).
Descriptive statistics for continuous variables are expressed
as mean±standard deviation. And categorical variables are ex-
pressed as frequency and percentage. All variables were test-
ed for normal distributions using the Kolmogorov-Smirnov
test. The Student t-test was applied to compare means of con-
tinuous variables. Categorical data were tested using the χ2
test or Fisher’s extract test. Statistical significance were con-
sidered if p-values were less than 0.05 (p＜0.05). Univariate
analysis and linear regression were performed. This study is
approved by the institutional review board of Seoul National
University Bundang Hospital, Seongnam, Korea.
Venoarterial ECMO was applied in 22 (84.6%) and veno-
venous in 4 cases. The reasons for applications of ECMO
were cardiac arrest in 12, cardiogenic shock in 9, respiratory
failure in 4, and septic shock in 1. Ventricular tachycardia
was classified as cardiac arrest. Patients with myocarditis
were 4, who presented with cardiogenic shock in 2 and car-
diac arrest in 2. During this period, left ventricular vent were
not done for myocarditis. Successful weaning from ECMO
was achieved in 15 cases (57.7%), and survival discharge rate
was 34.6% (9 cases) in this group. Overall mortality rate was
65.4%. In the weaning success group, there were 10 cases of
male patients and 5 of female. In the group of weaning fail-
ure, male was 5 and female was 6 cases. There were no stat-
istically difference in sex, age, body weight, hypertension,
diabetes, smoking, and preceding diseases for initiation of
ECMO, except myocarditis (Table 2). Mean duration of appli-
cation of ECMO was 111.39±54.06 hours (range, 48.22 to
267.97 hours). In univariate analysis, myocarditis (p=0.022)
was the only independent risk factors on weaning failure.
Hemoglobin level at 24 hours after initation of ECMO
showed equivocal significance of p-value 0.05. Changes in
platelet counts, level of AST, total bilirubin, glucose, and al-
Prognosis in Prolonged ECMO
− 239 −
Table 4. AUROC
AUROC±SE p-value95% CIAUROC±SEp-value 95% CI
Numerals indicate hours after initiation of ECMO (for example ECMO24Hb is hemoglobin level 24 hours after initiation of ECMO).
AUROC, area under receiver of characteristics; SE, standard error; CI, confidence interval; Adm, admission; Hb, hemoglobin;
ECMO, extracorporeal membrane oxygenation; Bwt, body weight; Plt, platelet counts; DurationHr, operating time of ECMO by
hour; Glu, glucosel; IS, Wernovsky inotropic score; Alb, albumin; TB, total bilirubin; AST, aspartate aminotransferase; VIS, vaso-
active inotropic score; Cr, creatinine; BE, base; AKIN, Acute Kidney Injury Network score; MAP, mean arterial pressure; SAPS2,
severity of disease classification system II; SOFA, sequential organ failure assessment.
Table 5. Patients died after successful weaning of ECMO
Number Sex/ageIndication of ECMO
weaning to death
Cause of death
Cardiogenic shock (myocarditis)
Recurrent cardiac arrest (R/O PTE)
Recurrent VT (AMI)
Cardiac arrest (STEMI)
Respiratory failure (interstitial pneumonia)
Considering lung TPL
Cardiogenic shock (STEMI)
2 hr 35 min
ARF, ARDS, sepsis
R/O recurrence of PTE
Sustained ventricular fibrillation
ECMO, extracorporeal membrane oxygenation; ARF, acute renal failure; ARDS, acute respiratory distress; R/O, rule out; PTE, pul-
monary thromboembolism; VT, ventricular tachycardia; AMI, acute myocardial infarction; STEMI, ST elevation myocardial infarction;
DIC, disseminated intravascular coagulation; TPL, transplantation.
bumin during ECMO were statistically insignificant. Further,
mean arterial pressure and the amount of blood transfusion
were not significant as well. Base excess, flow of ECMO, in-
otropic score, and changes in the level of serum creatinine
did not show any statistical differences. Eleven cases of suc-
cessful weaning groups are showed lower AKIN48-hour stage,
but no statistical significance (Table 3).
SAPS and SOFA scores were not matched with prognosis
of ECMO. There were no statistical significant factors except
myocarditis. We assumed that there were some hidden fac-
Tae-Hun Kim, et al
− 240 −
tors, which were showed no significant p-value due to small
number of total cases. So we applied area under receiver op-
erating characteristics curve, to find that factors. The level of
hemoglobin at 24 hours after ECMO, and base excess at 48
hours after ECMO were showed more than 0.7 (Table 4). If
the number of cases is enough, these factors may be consid-
ered as cut off values. Although successful weaning of
ECMO, 6 patients died. Brief explanation about them is pre-
sented as Table 5.
Due to myocardial destruction and remodeling, persistent
ventricular dysfunction and dilated cardiomyopathy occurs
frequently in myocarditis. Insufficient drainage of right ven-
tricular and bronchial circulation will lead to left ventricular
(LV) distension. LV distension interferes with myocardial rest
and thus greatly imposes impact on influencing the prognosis.
Therefore, LV venting may be helpful in patients with my-
ocarditis during ECMO. In Seoul National University Bun-
dang Hospital, myocarditis patients did not received LV vent-
ing for various reasons. So weaning failures from ECMO in
these patients might be result from this. Hemorrhage related
with the use of anticoagulants and the formation of thrombus
are typical complications that may be occur during ECMO.
Aside from this, long term management may cause severe
damage for organs and thus leading to poor prognosis.
Leukocytes, thrombocytes, and complements are activated by
ECMO. Acute inflammatory-like reaction can be introduced
during the first 24 to 48 hours by the secretion of cytokines,
which may cause capillary leakage and thus lead to intra-
vascular volume depletion [5,6]. In addition, it is known that
the long instillation of ECMO can cause problems, such as
hemolysis, and lead to organ damages. Lin et al.  reported
that the change in creatinine levels and hour urine during
ECMO is closely related to the prognosis. A few years later,
he reported that AKIN scores at the time of 48 hours after
initiation of ECMO was close related with the prognosis .
In this study, the relationship between AKIN score and
ECMO weaning was not statistically confirmed. However,
there were tendency of less severe degree of metabolic acido-
sis, slightly higher flow of ECMO, and less inotropic score
and changes in the level of serum creatinine in the successful
weaning group. It is a likely possibility that these results
arose due to small sample size of patients. If larger number
of ECMO patients, the outcome may differ.
Recently, new ECMO equipments are being introduced to
reduce the occurrence of hemolysis. Studies regarding these
devices and its impact on lowering kidney injuries may be
needed. The degree of acidosis 48 hours after initiation of
ECMO may reflect the degree of renal insufficiency. The he-
moglobin level at the time of 24 hours after initiation of
ECMO, regardless of the amounts of transfusion, may be the
possible prognostic factor, and degree of acidosis at 48 hours
after initiation of ECMO is also.
Maybe due to the small population of the sample size,
meaningful factors related with prognosis were not found in
this study. The increase in the duration of ECMO causes or-
gan dysfunction and will provoke a change in the hemoglobin
level. In addition, a progression of acidosis will adversely af-
fect the prognosis of the patients. From this perspective, the
result of this study may show some relationship between fac-
tors and prognosis. With reduction of hemolysis and acidosis,
the renal injury will occur less frequently and thus, will result
in a better prognosis of ECMO patients.
The first limitation of this present study was that number
of the patients enrolled in the study was too small to prove
or disprove the existing hypothesis. This could lead to se-
lection bias. Secondly, the comparison was not possible be-
tween the LV venting group in patients with myocarditis and
no venting. It is not possible to differentiate whether my-
ocarditis itself is a poor prognosis factor, or LV venting.
However, if the situation of myocarditis is considered, the
possibility of the latter is more likely [9,10].
This study showed that the hemoglobin levels of 24 hours
after ECMO and degrees of acidosis 48 after initiation of
ECMO are thought as prognostic factors. And patients with
myocarditis had poor prognosis in our cases. There was no
statistical relation between ECMO and renal injury. Further
study may be needed in the future.
Prognosis in Prolonged ECMO
− 241 −
1. Dalton HJ. Extracorporeal life support: moving at the speed
of light. Respir Care 2011;56:1445-53.
2. Yan X, Jia S, Meng X, et al. Acute kidney injury in adult
postcardiotomy patients with extracorporeal membrane oxy-
genation: evaluation of the RIFLE classification and the
Acute Kidney Injury Network criteria. Eur J Cardiothorac
3. Mehta RL, Kellum JA, Shah SV, et al. Acute Kidney Injury
Network: report of an initiative to improve outcomes in
acute kidney injury. Crit Care 2007;11:R31.
4. Le Gall JR, Lemeshow S, Saulnier F. A new Simplified Acute
Physiology Score (SAPS II) based on a European/North
American multicenter study. JAMA 1993;270:2957-63.
5. Godin C, Caprani A, Dufaux J, Flaud P. Interactions between
neutrophils and endothelial cells. J Cell Sci 1993;106(Pt
6. Haller H. Endothelial function: general considerations. Drugs
1997;53 Suppl 1:1-10.
7. Lin CY, Chen YC, Tsai FC, et al. RIFLE classification is
predictive of short-term prognosis in critically ill patients
with acute renal failure supported by extracorporeal mem-
brane oxygenation. Nephrol Dial Transplant 2006;21:2867-73.
8. Chen YC, Tsai FC, Chang CH, et al. Prognosis of patients
on extracorporeal membrane oxygenation: the impact of acute
kidney injury on mortality. Ann Thorac Surg 2011;91:137-42.
9. Jan SL, Lin SJ, Fu YC, et al. Extracorporeal life support
for treatment of children with enterovirus 71 infection-re-
lated cardiopulmonary failure. Intensive Care Med 2010;36:
10. Rogers JG, Milano CA. The role for mechanical circulatory
support for cardiogenic shock. In: Hochman J, Ohman EM.
Cardiogenic shock. Hoboken: Wiley-Blackwell Publishing;
2009. p. 209.