European Journal of Heart Failure (2018) 20, 572–581 RESEARCH ARTICLE

doi:10.1002/ejhf.958

Acute kidney injury in cardiogenic shock:

definitions, incidence, haemodynamic
alterations, and mortality
Tuukka Tarvasmäki1,2*, Mikko Haapio3, Alexandre Mebazaa4, Alessandro Sionis5,
José Silva-Cardoso6, Heli Tolppanen2,7, Matias Greve Lindholm8, Kari Pulkki9,
John Parissis10, Veli-Pekka Harjola1, and Johan Lassus2, on behalf of the CardShock
Study Investigators†
1 Emergency Medicine, University of Helsinki, and Department of Emergency Medicine and Services, Helsinki University Hospital, Helsinki, Finland; 2 Heart and Lung Center,
University of Helsinki and Helsinki University Hospital, Helsinki, Finland; 3 Abdominal Center, Nephrology, University of Helsinki and Helsinki University Hospital, Helsinki,
Finland; 4 INSERM U942, Hôpital Lariboisière, APHP and University Paris Diderot, Paris, France; 5 Intensive Cardiac Care Unit, Cardiology Department, Hospital de la Santa Creu i
Sant Pau, Biomedical Research Institute IIB Sant Pau, Universitat de Barcelona, Barcelona, Spain; 6 CINTESIS - Center for Health Technology and Services Research, Department of
Cardiology, Faculty of Medicine, University of Porto, São João Medical Center, Porto, Portugal; 7 Heart Center, Päijät-Häme Central Hospital, Lahti, Finland; 8 Intensive Cardiac
Care Unit, Rigshospitalet, Copenhagen University Hospital, Copenhagen, Denmark; 9 Department of Clinical Chemistry, University of Eastern Finland and Eastern Finland
Laboratory Centre, Kuopio, Finland; and 10 Heart Failure Unit, Attikon University Hospital, Athens, Greece

Received 10 April 2017; revised 5 June 2017; accepted 27 June 2017 ; online publish-ahead-of-print 27 September 2017

Aims To investigate the incidence, haemodynamic alterations and 90-day mortality of acute kidney injury (AKI) in patients
with cardiogenic shock. We assessed the utility of creatinine, urine output (UO) and cystatin C (CysC) definitions
of AKI in prognostication.
.....................................................................................................................................................................
Methods Cardiogenic shock patients with serial plasma samples (n = 154) from the prospective multicenter CardShock study
and results were included in the analysis. Acute kidney injury was defined and staged according to the KDIGO criteria by
creatinine (AKIcrea ) and/or UO (AKIUO ). CysC-based AKI (AKICysC ) was defined similarly to AKIcrea . Changes in
haemodynamic parameters were assessed over time from baseline until 96 h. Mean age of the study population was
66 ± 12 years and 74% were men. Median baseline creatinine was 1.12 [interquartile range (IQR) 0.87–1.54] mg/dL
and CysC 1.19 (IQR 0.90–1.69) mg/L. The 90-day mortality was 38%. The incidences for AKI were: AKIcrea 31%,
AKIUO 50%, and AKICysc 33%. AKIcrea [odds ratio (OR) 12.2, 95% confidence interval (CI) 4.1–36.0] and AKICysC (OR
2.5, 95% CI 1.1–6.1), but not AKIUO , were independent predictors of mortality. However, a stricter UO cut-off of
<0.3 mL/kg/h for 6 h was independently associated with 90-day mortality (OR 3.6, 95% CI 1.4–9.3). Development of
AKI was associated with persistently elevated central venous pressure and decreased cardiac index and mean arterial
pressure.
.....................................................................................................................................................................
Conclusions Acute kidney injury is frequent in patients with cardiogenic shock and especially AKIcrea predicts poor outcome. The
KDIGO UO criterion seems, however, rather liberal and a stricter AKI definition of UO <0.3 mL/kg/h for at least
6 h seems more useful for mortality risk prediction. Haemodynamic alterations reflecting venous congestion and
hypoperfusion were associated with AKI.
..........................................................................................................
Keywords Cardiogenic shock • Acute kidney injury • KDIGO • Urine output • Haemodynamics •
Mortality

*Corresponding author. Emergency Medicine, University of Helsinki and Department of Emergency Medicine and Services, Helsinki University Hospital, PO Box 340, 00029 HUS,
Helsinki, Finland. Tel: +358 40 534 3955, Fax: +358 9 471 71488, Email: tuukka.tarvasmaki@fimnet.fi
† For the CardShock Steering Committee and list of Investigators see the Supplementary material online, Appendix S1.

© 2017 The Authors

European Journal of Heart Failure © 2017 European Society of Cardiology
Acute kidney injury in cardiogenic shock 573

Introduction the detection of shock. The exclusion criteria were CS caused by ongo-

........................................................................................................................................................................
ing haemodynamically significant arrhythmia and shock after cardiac or
Acute kidney injury (AKI) is characterized by an abrupt decrease non-cardiac surgery. The distribution of specific aetiologies of CS in
in glomerular filtration rate and/or urine output (UO), and fre- the study population has been previously described by Harjola et al.18
quent among the critically ill, including patients with acute myocar-
dial infarction and acute heart failure (AHF).1 The Kidney Dis-
ease: Improving Global Outcomes (KDIGO) definition of AKI has
Data collection, plasma sampling
replaced the previous RIFLE and AKIN criteria.2 – 5 The current and acute kidney injury definitions
AKI definition is based on an increase in plasma creatinine and/or Detailed medical history and patient characteristics were recorded. In
a decrease in UO, and is divided into stages 1–3 depending on addition, medical and invasive treatment procedures were registered.
severity.4 While even small increases in creatinine levels have been Urine output was recorded at 6, 12, 18, and 24 h; an average UO was
shown to be associated with poor prognosis, the UO criteria calculated for consequent 6 h intervals dividing the total 6 h UO by
have not been extensively validated,4 and conflicting results have body weight and time. Cardiac index (CI) was measured from patients
with a pulmonary artery catheter (n = 61), and central venous pressure
been reported on whether the threshold of <0.5 mL/kg/h for 6
(CVP) from those with either a pulmonary artery catheter or a central
h is adequate for assessment of prognosis.6 – 8 In addition, cys-
venous catheter (n = 100). These data were collected at baseline (0 h),
tatin C (CysC), an alternative marker of kidney function and risk and 6, 12, 18, 24, 36, 48, 72 and 96 h; the mean values of measurements
stratification,9 has been suggested to possess prognostic value both between 0–12 h, 18–24 h, 36–48 h, and 72–96 h were calculated for
in AHF and AKI.10 – 12 further analyses.
Cardiogenic shock (CS) is a cardiac emergency and the most Serial plasma samples were collected at study baseline (0 h), and
severe form of AHF.13 ST-elevation myocardial infarction is the at 12, 24, 36 and 48 h. Of 178 patients with baseline samples, 154
most common cause of CS, but the aetiological spectrum is had one or more additional samples taken during the first 48 h after
broad and many other conditions may also lead to CS. Regardless baseline and these constitute the study population; a flowchart is
of aetiology, CS is characterized by inadequate cardiac output included in the Supplementary material online, Appendix S1. The blood
leading to hypotension and signs and/or symptoms of end-organ samples were centrifuged, plasma was separated and immediately
stored frozen at −80 ∘ C until analysed. Creatinine (Roche Diagnostics,
hypoperfusion such as oliguria.14 While venous congestion and
Basel, Switzerland) and CysC (Abbott Laboratories, Abbott Park, IL,
reduced renal perfusion are considered the main mechanisms for
USA) were analysed centrally. Creatinine levels were measured with
impaired and worsening renal function in heart failure,15 little is Cobas 6000 automated analyzer (Roche Diagnostics) and CysC levels
known about haemodynamic parameters, their evolution over time with Architect ci8200 automated analyzer (Abbott Laboratories).
and relation to AKI in CS. The overall incidence, risk factors and We defined and staged AKI similarly to the KDIGO definition
mortality of AKI patients with critical illness, such as sepsis, are and staging (see Supplementary material online, Appendix S1).4 For
well characterized.1,16,17 In contrast, data on AKI in the setting of creatinine-based AKI (AKIcrea ) ≥0.3 mg/dL or ≥50% increase in cre-
CS are very scarce, reflecting a gap in knowledge. atinine from baseline value was defined as stage 1, 100–199% increase
The aim of this study was to describe the incidence of AKI using as stage 2, and ≥200% increase or increase to ≥4.0 mg/dL in creati-
the contemporary KDIGO creatinine and UO criteria in a multi- nine as stage 3 AKI. For AKIUO , UO <0.5 mL/kg/h for at least 6 h was
national prospective observational study on CS. In addition, we defined as stage 1, <0.5 mL/kg/h for at least 12 h as stage 2, and <0.3
mL/kg/h for at least 24 h or anuria (defined as UO <50 mL/12 h or
investigated the utility of CysC-based definitions of AKI (AKICysC ).
<100 mL/24 h) for at least 12 h as stage 3 AKI. Assessment of AKIcrea
The associations of AKI with haemodynamic alterations and 90-day
was based on creatinine levels from baseline until 48 h, and the highest
mortality were assessed. Finally, we explored a previously sug- increase within this time was used for staging. AKIUO was staged by
gested stricter definition for UO-based AKI (AKIUO ) for mortality the lowest UO for a time interval within the first 24 h. For AKICysC , a
prediction.7 similar definition was used as for AKIcrea : ≥0.3 mg/L or ≥50% increase
from baseline as stage 1, 100–199% increase as stage 2, and ≥200%
increase as stage 3 AKICysC .11,12,19,20 Renal replacement therapy (RRT)
Methods was not included in AKI stage 3 criteria unless stated otherwise.
The primary endpoint of interest was 90-day all-cause mortality.
Study population Two patients had missing follow-up data after hospital discharge. The
The CardShock study (ClinicalTrials.gov identifier: NCT01374867) is CardShock study was approved by local ethics committees at the
a prospective multicentre study, which recruited patients with CS at participating centres (see Supplementary material online for details,
nine hospitals in eight European countries (Czech Republic, Denmark, Appendix S1) and conducted in accordance with the Declaration of
Finland, Greece, Italy, Poland, Portugal, and Spain) between October Helsinki.
2010 and December 2012.18
In addition to acute cardiac cause, the inclusion criteria were: sys-
tolic blood pressure (SBP) <90 mmHg for 30 min or need for vaso- Statistical analyses
pressor to maintain SBP >90 mmHg (in the absence of hypovolaemia or Results are presented as numbers and percentages for categorical
after adequate fluid challenge), and signs of organ hypoperfusion (any of variables, the mean with standard deviation or the median with
the following: cold extremities, confusion or altered mental status, olig- interquartile range for continuous variables, as appropriate. Group
uria during the previous 6 h, or blood lactate >2 mmol/L). Patients had comparisons were performed with Fisher’s exact or chi-square test
to be aged over 18 years and included in the study within the first 6 h of for categorical, Mantel–Haenszel trend test for ordinal, and t-test

© 2017 The Authors

European Journal of Heart Failure © 2017 European Society of Cardiology
574 T. Tarvasmäki et al.

or Mann–Whitney U test for continuous variables, as appropriate. P < 0.001] at baseline; the OR for death at 90 days was 1.75 (95%

........................................................................................................................................................................
We examined the 90-day survival with Kaplan–Meier curves and CI 1.11–2.77) per mg/dL increase for creatinine, and 1.03 (95%
performed group comparisons with log-rank test. For assessment CI 1.01–1.04) per mL/min/1.73m2 decrease in eGFR. All AKIcrea
of independent predictors of AKI, we used forward and backward stages were associated with higher 90-day mortality (Figure 2a).
selection of variables in logistic regression to calculate likelihood ratios,
After adjustment for different variable sets, AKIcrea was consis-
with significance <0.05 for inclusion and >0.1 for elimination; the final
tently associated with increased 90-day mortality (OR 12.2, 95%
model was further adjusted for age and estimated glomerular filtration
rate (eGFR) at baseline. We tested the independent association of
CI 4.1–36.0, for fully adjusted model) (Table 2). Including RRT as
AKI with 90-day mortality with logistic regression; multiple analyses part of the AKI stage 3 criteria did not markedly change the results
were performed with different variable sets including CardShock risk (data not shown).
score (see Supplementary material online, Appendix S1, for details)
as a continuous variable18 as well as gender and baseline SBP. The
additive value of a variable in mortality prediction was assessed via Urine output and mortality
the likelihood ratio test for nested models. Discriminative capability
AKIUO was not independently associated with death at 90 days
was assessed by the area under receiver operating curve (AUC). AUC
(Table 2) and had no additive value in 90-day mortality predic-
comparisons were performed with DeLong method. We used linear
mixed modelling to assess differences in haemodynamic measurements tion compared with AKIcrea alone (𝜒 2 = 0.843, P = 0.4 for com-
between groups over time. parison of nested models). AKIUO stages 2 and 3 were asso-
Odds ratios (ORs) are shown with 95% confidence intervals (CIs). ciated with increased rates of death (Figure 2b) and remained
We considered P-values <0.05 as statistically significant. We performed independently predictive of poor outcome after adjustment for
statistical analyses with IBM SPSS Statistics, version 24.0 (IBM Corp, gender, SBP and the CardShock risk score. However, stage 1
Armonk, NY, USA), and MedCalc Statistical Software, version 17.1 AKIUO (<0.5 mL/kg/h for 6 h) was not related to higher mortality
(MedCalc Software bvba, Ostend, Belgium). Venn diagram was drawn (Figure 2b).
with eulerAPE. We explored a stricter UO threshold (i.e. UO <0.3 mL/kg/h
for 6 h) previously used in the critically ill and found it to perform
better in predicting 90-day mortality. This threshold showed better
Results discriminative capability for death at 90 days than <0.5 mL/kg/h for
Baseline characteristics of the study patients (n = 154) are shown 6 h, and combining UO <0.3 mL/kg/h 6 h with AKIcrea improved
in Table 1. In brief, mean age was 66 ± 12 years and 74% were men. 90-day mortality prediction compared with AKIcrea alone (Figure
Median creatinine was 1.12 (0.87–1.54) mg/dL, eGFR 64 (43–87) 2d and Supplementary material online, Table S2; 𝜒 2 = 7.652, P =
mL/min/1.73 m2 and CysC 1.19 (0.90–1.69) mg/L. Acute coronary 0.006 for comparison of nested models). AKIUO with the stricter
syndrome was the aetiology of CS in 81% of patients, and 76% threshold also retained an independent association with 90-day
underwent coronary angiography within 72 h before and 12 h after mortality after multivariable adjustment (OR 3.6, 95% CI 1.4–9.3)
baseline. All-cause 90-day mortality was 38%. (Table 2).
The incidences of AKI stages according to the different defini-
tions are shown in Figure 1a. Comparison of patient characteristics
between those with and without AKIcrea are shown in Table 1; for Factors associated with development
AKICysC and AKIUO , see the Supplementary material online, Table of AKIcrea and AKIUO
S1. Using the KDIGO criteria for creatinine and UO, 31% devel- Lower baseline arterial pH and previous use of diuretics were
oped AKIcrea and 50% AKIUO . AKICysC was observed in 33% of independently associated with higher incidence of AKIcrea whereas
patients. In patients who developed AKIcrea , 74% fulfilled the cri- lower baseline eGFR was an independent predictor of AKIUO .
teria during the first 24 h and 26% (12/47) between 24 to 48 h Whereas use of intravenous furosemide during the first 24 h cor-
from baseline. As shown in Figure 1b, the different definitions of related positively with cumulative 24 h UO, no association with
AKI were not entirely concordant; half of the AKIUO patients had AKIUO existed; instead, it was independently associated with higher
no AKIcrea or AKICysC . incidence of AKIcrea when added to the multivariable model of
Renal replacement therapy was initiated within the first 48 h baseline variables. The only baseline variable that independently
in 16 patients. Almost all of these patients (94%) had AKIUO , of predicted the development of AKIUO was lower eGFR. Previ-
which 80% had AKIUO stage ≥2, whereas 9 (56%) had AKIcrea and ous use of angiotensin-converting enzyme inhibitors/angiotensin
10 (63%) AKICysC . Continuous RRT was used in nine patients and II receptor blockers was not associated with increased inci-
seven received intermittent haemodialysis either with or without dence of AKI. Among those who underwent coronary angiogra-
ultrafiltration. Isolated ultrafiltration was used in five patients. phy within 72 h before and 24 h after baseline, the amount of
contrast media volume was positively associated with AKIUO , but
not AKIcrea .
Renal function, AKIcrea , and mortality Of note, use of dopamine within the first 24 h was not associated
Compared with patients alive at 90 days, decedents had higher cre- with development of AKI. In addition, the amount of administered
atinine [1.29 (0.97–1.83) vs. 1.04 (0.78–1.36) mg/dL, P = 0.001], fluid during the first 24 h correlated neither with the cumulative
and lower eGFR [51 (33–71) vs. 71 (53–96) mL/min/1.73 m2 , 24 h UO nor development of AKIUO .

© 2017 The Authors

European Journal of Heart Failure © 2017 European Society of Cardiology
Acute kidney injury in cardiogenic shock 575

Table 1 Patient characteristics and the differences between patients with and without AKIcrea

Characteristics Total No AKIcrea AKIcrea P-value

(n = 154) (n = 107) (n = 47)
...........................................................................................................................................
Age (years) 66 ± 12 65 ± 12 69 ± 12 0.049
Female gender 26% 25% 28% 0.8
Medical history
Ischaemic heart disease 33% 30% 40% 0.2
Myocardial infarction 25% 21% 32% 0.17
Percutaneous coronary intervention 17% 21% 9% 0.07
Coronary artery bypass graft surgery 6% 4% 13% 0.036
Chronic heart failure 16% 13% 19% 0.4
Hypertension 62% 60% 68% 0.3
Diabetes 28% 28% 28% >0.9
Renal insufficiency 11% 7% 19% 0.033
Medication
ACEi/ARB 41% 43% 36% 0.4
Mineralocorticoid receptor antagonist 6% 4% 11% 0.13
Diuretic 28% 24% 38% 0.06
Baseline characteristics
Resuscitation prior to inclusion 29% 27% 34% 0.4
ACS as CS aetiology 81% 78% 87% 0.16
Confusion 68% 66% 72% 0.4
Oliguria >6 h prior to inclusion 52% 45% 70% 0.005
Systolic blood pressure (mmHg) 78 ± 14 78 ± 15 78 ± 12 0.9
Sinus rhythm 79% 85% 66% 0.007
Left ventricular ejection fraction (%) 33 ± 13 34 ± 14 31 ± 12 0.2
Biochemistry
Haemoglobin (g/L) 130 ± 24 130 ± 22 129 ± 27 >0.9
Sodium (mmol/L) 136 ± 5 136 ± 6 137 ± 4 0.6
Creatinine (mg/dL) 1.12 (0.87–1.54) 1.06 (0.79–1.48) 1.22 (0.96–1.60) 0.018
Cystatin C (mg/L) 1.19 (0.90–1.69) 1.13 (0.85–1.51) 1.61 (1.09–1.92) 0.001
eGFRCKD-EPI (mL/min/1.73 m2 ) 64 (43–87) 70 (45–95) 56 (35–74) 0.007
Lactate (mmol/L) 2.6 (1.6–5.3) 2.4 (1.4–4.2) 3.7 (2.3–7.0) 0.001
Arterial pH 7.31 ± 0.13 7.33 ± 0.13 7.27 ± 0.12 0.019
Arterial pH <7.3 48% 42% 62% 0.027
hsTnT (ng/L) 2275 (379–5985) 1828 (350–5784) 2862 (470–8849) 0.2
NT-proBNP (ng/L) 2247 (550–8253) 2475 (511–7487) 1974 (669–9716) 0.7
Treatment/procedure
Angiography 83% 85% 79% 0.3
Before study inclusion 62% 63% 60% 0.7
Within 72 h before and 12 h after inclusion 76% 76% 77% >0.9
Contrast volume (mL) 180 (120–250) 160 (110–246) 200 (148–300) 0.032
I.v. furosemide within first 24 h 64% 56% 81% 0.003
Cumulative dose (mg) 100 (40–235) 80 (23–173) 120 (50–470) 0.07
Vasopressor 83% 79% 94% 0.021
Intra-aortic balloon pump 56% 50% 72% 0.009
LVAD or ECMO 4% 1% 11% 0.004
Outcome
RRT 14% 8% 28% 0.002
90-day mortality 38% 24% 70% <0.001
RRT or death at 90 days 42% 28% 74% <0.001

Data are expressed as proportion of patients (%), means ± standard deviation, or medians (interquartile range), as appropriate.
ACEi, angiotensin-converting enzyme inhibitor; AKIcrea , acute kidney injury by creatinine criteria of the KDIGO guideline; ACS, acute coronary syndrome; ARB, angiotensin II
receptor blocker; CS, cardiogenic shock; ECMO, extracorporeal membrane oxygenation; eGFRCKD-EPI , estimated glomerular filtration rate with the Chronic Kidney Disease
Epidemiology Collaboration equation; hsTnT, high-sensitivity troponin T; I.v., intravenous; LVAD, left ventricular assist device; NT-proBNP, N-terminal pro-B-type natriuretic
peptide; RRT, renal replacement therapy (during hospitalization).

© 2017 The Authors

European Journal of Heart Failure © 2017 European Society of Cardiology
576 T. Tarvasmäki et al.

Figure 1 (a) Incidences of acute kidney injury (AKI) stages according to the creatinine, urine output and cystatin C definitions, and (b) the
AKI definitions and their overlapping proportions shown as a Venn diagram. AKIcrea , acute kidney injury by creatinine criteria of the KDIGO
guideline; AKIUO , acute kidney injury by urine output criteria of the KDIGO guideline; AKICysC , acute kidney injury by cystatin C criteria,
similar to the creatinine criteria of the KDIGO guideline.

Table 2 Unadjusted and adjusted associations of the acute kidney injury definitions with 90-day mortality

Adjustment model AKIcrea AKIUO UO < 0.3 mL/kg/h for 6 h

............................................................. ....................................
OR (95% CI) P-value OR (95% CI) P-value OR (95% CI) P-value
...........................................................................................................................................
Unadjusted 7.5 (3.5–12.3) <0.001 2.1 (1.05–4.0) 0.035 4.7 (2.2–9.8) 0.001
Model 1 7.3 (3.3–16.4) <0.001 1.7 (0.8–3.4) 0.15 3.7 (1.7–8.0) 0.001
Model 2 7.5 (3.2–17.8) <0.001 1.6 (0.8–3.5) 0.2 3.9 (1.7–9.0) 0.001
Model 3 12.2 (4.1–36.0) <0.001 1.5 (0.6–3.5) 0.4 3.6 (1.4–9.3) 0.008

AKIcrea , acute kidney injury by creatinine criteria of the KDIGO guideline; AKIUO , acute kidney injury by urine output criteria of the KDIGO guideline; CI, confidence interval;
OR, odds ratio, UO = urine output.
Model 1: adjusted for age, gender.
Model 2: adjusted for age, gender, systolic blood pressure, estimated glomerular filtration rate.
Model 3: adjusted for systolic blood pressure, gender, CardShock risk score (as a continuous variable; includes age and estimated glomerular filtration rate as variables).18

Haemodynamic alterations in relation AKICysC : definitions and association

...............................................

to the development of acute kidney with mortality

injury Cystatin C levels were higher in non-survivors than survivors [1.39
Figure 3 shows the incidences of AKIcrea in haemodynamic mea- (1.04–1.96) vs. 1.13 (0.81–1.53) mg/L, P = 0.001] and baseline
surement strata as well as the evolution of CVP, mean arterial CysC had an OR for 90-day mortality of 1.98 (95% CI 1.24–3.16,
pressure (MAP) and CI over time in patients with and with- P = 0.004) per mg/L increase. The incidence of AKICysC was
out AKI; for AKIUO , see the Supplementary material online, almost equal to and identified a similar population of patients as
Figure S1. The highest CVP and lowest CI and MAP cate- AKIcrea . The concordance with AKIUO was only moderate (Figure
gories were associated with the highest incidence of AKIcrea 1b). Most patients had stage 1 AKICysC (Figure 1a), and AKICysC
and AKIUO . Analogously, higher levels of CVP and lower MAP stages 2–3 were seen only in a few patients. Increasing AKICysC
and CI were associated with both AKIcrea and AKIUO (P < 0.5 stage was associated with a stepwise increase in mortality rates
for all pooled between-group comparisons). In addition, higher (Figure 2c). Furthermore, AKICysC showed an independent asso-
AKIcrea stage and lower UO were associated with higher CVP ciation with increased 90-day mortality after adjustment for gen-
and lower MAP and CI (see Supplementary material online, der, SBP and the CardShock risk score (OR 2.5, 95% CI 1.1–6.1,
Figure S2). P = 0.04).

© 2017 The Authors

European Journal of Heart Failure © 2017 European Society of Cardiology
Acute kidney injury in cardiogenic shock 577

Figure 2 Comparison of survival curves stratified by (a) acute kidney injury staging defined by creatinine (AKIcrea ), (b) AKI staging defined
by urine output (AKIUO ), (c) AKI staging defined by cystatin C (AKICysC ), and (d) AKIcrea and urine output cut-off of <0.3 mL/kg/h for 6 h
(UO<0.3 ). On the right side of each panel, significant differences in pairwise comparisons are highlighted with single asterisk (log-rank P < 0.05)
and double asterisk (log-rank P < 0.01). AKIUO stage comparison performed among study patients for at least 24 h (b).

Discussion in prognostication, we propose a more stringent cut-off of <0.3

.............................

mL/kg/h for 6 h, which appeared useful in mortality prediction.

This is the first study to prospectively evaluate AKI by contempo-
rary KDIGO definitions in patients with CS, and also to describe
AKI definitions using CysC. We found AKI to occur in more
than one-third of the patients and to be associated with poor
prognosis. However, the different AKI definitions were not entirely
concordant. The AKIcrea definition had a strong association with
increased mortality. In contrast, AKIUO was not independently
associated with mortality. To improve the UO-based definition
Acute kidney injury was associated with haemodynamic alterations
reflecting persistent venous congestion and hypoperfusion.
AKIcrea criteria and mortality prediction
While AKI in hospitalized patients or the critically ill has been well
studied, there is insufficient knowledge on AKI in the setting of
CS. There are no previous data on current KDIGO AKI criteria

© 2017 The Authors

European Journal of Heart Failure © 2017 European Society of Cardiology
578 T. Tarvasmäki et al.

Figure 3 Incidence of acute kidney injury defined by KDIGO creatinine criteria (AKIcrea ) in haemodynamic measurement strata (upper panels),
and the evolution of haemodynamic alterations over time (lower panels) in no-AKIcrea and AKIcrea patients. In the upper panels, P-values for
trend for the association between AKI incidence and haemodynamic measurement strata are shown in each time interval. In the lower panels,
lines represent the mean values with 95% confidence intervals of haemodynamic measurements within the time intervals. In the lower panels,
P < 0.05 for all between-group comparisons and P = NS for all time-by-group interactions. CI, cardiac index; CVP, central venous pressure;
MAP, mean arterial pressure.

in patients with CS. The most recent report, a single-centre detection and AKIUO to be more frequent than AKI by creatinine
..................................................................

substudy of the IABP-SHOCK II trial, used only stage 1 creatinine criteria.24 – 26 Data on the utility and concordance of the current
criteria for definition of AKI.21 Two other single-centre studies KDIGO definitions of AKIUO are, however, conflicting. While the
on AKI in CS had specific inclusion and/or exclusion criteria and FINNAKI study including Finnish ICU patients showed severe
used heterogeneous definitions of AKI.22,23 Thus, findings in the oliguria to be highly predictive of AKIcrea ,8 some studies have
previous studies of CS are not directly comparable with those shown that many patients with oliguria do not develop AKIcrea .26,27
of our study. However, our study confirmed two key messages: The concordance in the frequency and staging of AKI between
the high incidence of AKIcrea in CS and its association with a the definitions was even weaker in our study; a rather large
notably poor prognosis. The results of our study are also similar proportion of patients with AKIUO did not develop AKIcrea and
to other studies using KDIGO definitions in unselected critically ill vice versa. Consequently, the current UO threshold in the KDIGO
patient populations.6,16,17 Nevertheless, it must be emphasized that guidelines had a modest association with development of AKIcrea .
mortality in CS patients is consistently higher than in the overall All in all, concordance between AKIUO and AKIcrea seems to be
unselected intensive care unit (ICU) population. Mortality rates inconsistent.8,26 – 28
in CS are further increased by AKIcrea , which was a strong and In our study, only stage ≥2 AKIUO , i.e. a longer duration (at least
independent predictor of death at 90 days. 12 h) of UO <0.5 mL/kg/h, was independently associated with
increased mortality. Although some studies have found the UO
criteria to perform well in mortality prediction, our data do not
Urine output as a criterion for acute fully support this view. For the 6 h measurement period, we tested
kidney injury and in mortality prediction a previously suggested cut-off of <0.3 mL/kg/h and found it to be
In this study, half of the patients fulfilled the AKIUO criteria more appropriate for mortality prediction7 . This cut-off was also
compared with only one-third having AKIcrea . The UO criterion more specific for and had a stronger association with incidence
has previously been found to increase the sensitivity for AKI of AKIcrea . In addition, we found that combining this cut-off with

© 2017 The Authors

European Journal of Heart Failure © 2017 European Society of Cardiology
Acute kidney injury in cardiogenic shock 579

AKIcrea provided additive value in mortality prediction and patients time points than creatinine,10 and to be at least as accurate or

........................................................................................................................................................................
fulfilling both AKIcrea and UO <0.3 mL/kg/h for 6 h criteria had the even superior in estimation to acute changes in eGFR,33 and
highest mortality. the independence of CysC level from height, gender, age, and
Initial resuscitation with large amounts of fluids in the critically muscle mass is advantageous. CysC-based AKI definition might
ill and diuretic use may obscure the detection of AKIUO by affecting be particularly useful in the elderly and in patients with changes
diuresis. Whereas intravenous furosemide positively correlated in creatinine metabolism or muscle mass due to catabolic critical
with urinary volume, no association with AKIUO existed. Instead, illness for example. A recent study in an unselected ICU pop-
intravenous furosemide was associated with increased incidence ulation reported that AKICysC was slightly better in short-term
of AKIcrea . Furthermore, the amount of administered fluid did not mortality prediction than creatinine-based definitions of RIFLE,
correlate with urinary volume. As UO is a vital component in AKIN, or KDIGO.12 While our study showed that AKICysC was
monitoring kidney function and an easily measurable parameter, independently associated with mortality, it provided no clear
and despite its shortcomings, its inclusion instead of omission in advantage over AKIcrea . Although the incidence of AKI was similar
the AKI definition is justifiable and the stricter UO cut-off for AKI using CysC and creatinine definitions, the two markers identified
detection seems preferable in CS. AKI patient cohorts with differences in AKI stages, which is in
line with a previous study on AHF patients.19 While the cut-off
Haemodynamic alterations in relation of 0.3 mg/L increase has been used in several cohorts to define
AKICysC ,19,20,32 definitions for AKICysC stages 2 and 3 have not been
to AKIcrea and urine output
well established. All in all, CysC seems to be a feasible alternative
This is the first study, to our knowledge, to assess serial haemody- as an AKI marker but further studies are warranted.
namic measurements in CS in relation to AKI and UO in particular.
Not only persistent venous congestion, as measured by an increase
in CVP, but also arterial and organ hypoperfusion reflected by a Limitations
lower arterial pressure and CI were shown to associate with both
There are some limitations that should be acknowledged. First,
AKI incidence and severity. Indeed, several studies have empha-
we assessed AKI only for the first 48 h. However, most of AKI
sized the role of venous congestion in the development of worsen-
in the critically ill occur within the first day,8 and also in our
ing renal function in cardiovascular diseases.29,30 Then again, while
study 74% of patients developed AKI within the first 24 h. Fur-
low cardiac output has been shown to associate with worsen-
thermore, many AKI patients died already within 48 h. Second,
ing renal failure in many settings, studies by Mullens et al.30 and,
UO was not recorded hourly but with fixed time intervals of 6 h.
more recently, Hanberg et al.31 have, however, shown that the role
However, measuring UO in fixed intervals has been suggested to
of CI in AHF is limited. The effect of hypoperfusion caused by a
be more sensitive and more practical than assessing consecutive
severe derangement in cardiac output and low arterial pressure
hourly measurements facilitating the application of the criteria.25
exceeding the autoregulation capacity of the kidney has proba-
Third, no data on creatinine prior to index hospitalization for CS
bly a more relevant role in CS than in less severe forms of AHF.
was available. On the other hand, the European Renal Best Prac-
The backward failure and severe neurohormonal activation caused
tice position statement, for example, recommends using baseline
by CS can lead to a significant increase in venous congestion as
rather than historical creatinine values or a calculated value based
well; venous congestion is further aggravated by fluid administra-
on an assumed GFR of 75 mL/min.5 No information on diabetes
tion used to counter hypotension and hypoperfusion. In addition to
type (I or II), time since diagnosis, or need for insulin treatment was
the haemodynamic derangements, the pronounced inflammatory
available. Finally, the assessment and detection of kidney function
response and neurohormonal activation further exacerbate renal
with plasma and urine biomarkers were not necessarily obtained
injury and dysfunction. Furthermore, use of nephrotoxic agents,
simultaneously with haemodynamic measurements and derange-
such as contrast media and diuretics, in remarkably labile condi-
ments. However, the changes in kidney function and haemodynamic
tions may produce excessive deleterious effects.
parameters occur in a continuum over time rather than at a par-
In addition to AKIcrea , also AKIUO was associated with such
ticular time point. Our results show a marked association between
haemodynamic alterations. However, the effect was more pro-
haemodynamic derangements persisting from before the detection
nounced in or seemingly driven by a more significant reduction in
of AKI until its detection and even thereafter.
UO as reflected by the difference between groups 0.3–0.5 and <0.3
mL/kg/h for 6 h, thus advocating for the assimilation of a stricter
cut-off to be used as AKIUO definition in CS.
Conclusion
Acute kidney injury is frequent in CS, and AKIcrea is strongly
Cystatin C as a marker of acute kidney associated with increased 90-day mortality. The current UO-based
injury KDIGO AKI definition was rather liberal compared with AKIcrea
Serum CysC, an alternative to creatinine as a filtration marker and not independently associated with increased mortality. Based
and a risk stratification tool,9 has been studied for detection on the current study, we propose a more stringent threshold,
and prognostication of AKI in AHF, critically ill, and surgical <0.3 mL/kg/h for 6 h, to improve mortality prediction. AKIcrea and
patients.10,11,19,32 It has been shown to predict AKI at earlier AKIUO were associated with haemodynamic alterations reflecting

© 2017 The Authors

European Journal of Heart Failure © 2017 European Society of Cardiology
580 T. Tarvasmäki et al.

venous congestion and hypoperfusion. Cystatin C may be used 2. Bellomo R, Ronco C, Kellum JA, Mehta RL, Palevsky P; Acute Dialysis Quality

........................................................................................................................................................................
Initiative Workgroup. Acute renal failure—definition, outcome measures, animal
as an alternative to creatinine for AKI definition and mortality
models, fluid therapy and information technology needs: the Second International
prediction in CS. Consensus Conference of the Acute Dialysis Quality Initiative (ADQI) Group. Crit
Care 2004;8:R204–R212.
3. Mehta RL, Kellum JA, Shah SV, Molitoris BA, Ronco C, Warnock DG, Levin
Supplementary Information A; Acute Kidney Injury Network. Acute Kidney Injury Network: report of an
initiative to improve outcomes in acute kidney injury. Crit Care 2007;11:R31.
Additional Supporting Information may be found in the online 4. Kidney Disease: Improving Global Outcomes (KDIGO) Acute Kidney Injury
Work Group. KDIGO Clinical Practice Guideline for Acute Kidney Injury. Kidney
version of this article: Int Suppl 2012;2:1–138.
Appendix S1. Definitions of acute kidney injury and staging; 5. Fliser D, Laville M, Covic A, Fouque D, Vanholder R, Juillard L, Van Biesen W;
flowchart of patients included in the study; CardShock risk score; Ad-hoc Working Group of ERBP. A European Renal Best Practice (ERBP) posi-
tion statement on the Kidney Disease Improving Global Outcomes (KDIGO)
and CardShock Steering Committee and Investigators. clinical practice guidelines on acute kidney injury: part 1: definitions, conser-
Table S1. Comparison of patient characteristics between no-AKI vative management and contrast-induced nephropathy. Nephrol Dial Transplant
and AKI defined either by urine output or cystatin C criteria. 2012;27:4263–4272.
6. Mandelbaum T, Scott DJ, Lee J, Mark RG, Malhotra A, Waikar SS, Howell MD,
Table S2. Discriminative capabilities of different thresholds for an Talmor D. Outcome of critically ill patients with acute kidney injury using the
average urine output for a 6 h interval. Acute Kidney Injury Network criteria. Crit Care Med 2011;39:2659–2664.
7. Md Ralib A, Pickering JW, Shaw GM, Endre ZH. The urine output definition of
Figure S1. Incidence of AKIUO in haemodynamic measurement
acute kidney injury is too liberal. Crit Care 2013;17:R112.
strata, and the evolution of haemodynamic alterations over time in 8. Vaara ST, Parviainen I, Pettila V, Nisula S, Inkinen O, Uusaro A; FINNAKI Study
no-AKIUO and AKIUO patients. Group. Association of oliguria with the development of acute kidney injury in the
critically ill. Kidney Int 2016;89:200–208.
Figure S2. Evolution of haemodynamic parameters over time
9. Shlipak MG, Matsushita K, Arnlov J, Inker LA, Katz R, Polkinghorne KR, Rothen-
stratified by AKIcrea stage, and urine output with cut-offs of 0.5 bacher D, Sarnak MJ, Astor BC, Coresh J, Levey AS, Gansevoort RT; CKD
mL/kg/h and 0.3 mL/kg/h. Prognosis Consortium. Cystatin C versus creatinine in determining risk based
on kidney function. N Engl J Med 2013;369:932–943.
10. Lassus J, Harjola VP. Cystatin C: a step forward in assessing kidney function and
Acknowledgements cardiovascular risk. Heart Fail Rev 2012;17:251–261.
11. Zhang Z, Lu B, Sheng X, Jin N. Cystatin C in prediction of acute kidney injury: a
The study was conducted in collaboration with the Global systemic review and meta-analysis. Am J Kidney Dis 2011;58:356–365.
REsearch on Acute Conditions Team (GREAT) network. 12. Zhou J, Liu Y, Tang Y, Liu F, Zhang L, Zeng X, Feng Y, Tao Y, Yang L, Fu P. A
comparison of RIFLE, AKIN, KDIGO, and Cys-C criteria for the definition of
Laboratory kits for analysis of NT-proBNP and hsTnT were
acute kidney injury in critically ill patients. Int Urol Nephrol 2016;48:125–132.
kindly provided by Roche Diagnostics, Basel, Switzerland. 13. Topalian S, Ginsberg F, Parrillo JE. Cardiogenic shock. Crit Care Med 2008;36(1
Suppl):S66–S74.
Funding 14. Reyentovich A, Barghash MH, Hochman JS. Management of refractory cardio-
genic shock. Nat Rev Cardiol 2016;13:481–492.
T.T. has received a personal scholarship from the Department of 15. Damman K, Testani JM. The kidney in heart failure: an update. Eur Heart J
2015;36:1437–1444.
Emergence Medicine and Services of Helsinki University Hospital,
16. Hoste EA, Bagshaw SM, Bellomo R, Cely CM, Colman R, Cruz DN, Edipidis K,
and personal grants from Ida Montin Foundation, Orion Research Forni LG, Gomersall CD, Govil D, Honore PM, Joannes-Boyau O, Joannidis M,
Foundation and Aarne Koskelo Foundation. J.L. has received Korhonen AM, Lavrentieva A, Mehta RL, Palevsky P, Roessler E, Ronco C, Uchino
S, Vazquez JA, Vidal Andrade E, Webb S, Kellum JA. Epidemiology of acute kidney
personal grants from Finska Läkaresällskapet, Paavo Nurmi Foun-
injury in critically ill patients: the multinational AKI-EPI study. Intensive Care Med
dation and W.E. Stockmann Foundation. The CardShock study was 2015;41:1411–1423.
supported by grants from Aarne Koskelo Foundation, the Finnish 17. Nisula S, Kaukonen KM, Vaara ST, Korhonen AM, Poukkanen M, Karlsson S,
Haapio M, Inkinen O, Parviainen I, Suojaranta-Ylinen R, Laurila JJ, Tenhunen J,
Cardiac Foundation, and the State funding for university-level
Reinikainen M, Ala-Kokko T, Ruokonen E, Kuitunen A, Pettila V; FINNAKI Study
health research. Group. Incidence, risk factors and 90-day mortality of patients with acute kidney
Conflict of interest: A.M. has received speaker’s honoraria from injury in Finnish intensive care units: the FINNAKI study. Intensive Care Med
2013;39:420–428.
The Medicines Company, Novartis, Orion, Roche, and Servier,
18. Harjola VP, Lassus J, Sionis A, Kober L, Tarvasmaki T, Spinar J, Parissis J,
and received a fee as member of the advisory board and/or Banaszewski M, Silva-Cardoso J, Carubelli V, Di Somma S, Tolppanen H, Zeymer
steering committee from Cardiorentis, The Medicines Company, U, Thiele H, Nieminen MS, Mebazaa A; CardShock Study Investigators and the
GREAT Network. Clinical picture and risk prediction of short-term mortality in
Adrenomed, MyCartis, ZS Pharma, and Critical Diagnostics. J.P.
cardiogenic shock. Eur J Heart Fail 2015;17:501–509.
received honoraria from Novartis, Orion Pharma and Roche Diag- 19. Lassus JP, Nieminen MS, Peuhkurinen K, Pulkki K, Siirila-Waris K, Sund R, Harjola
nostics. V.-P.H. has served on advisory boards for Bayer, BMS/Pfizer, VP; FINN-AKVA Study Group. Markers of renal function and acute kidney injury
in acute heart failure: definitions and impact on outcomes of the cardiorenal
Boehringer Ingelheim, MSD, Novartis, and Roche Diagnostics, and syndrome. Eur Heart J 2010;31:2791–2798.
received lecture fees from Bayer, BMS/Pfizer, and Orion Pharma. J.L. 20. Metra M, Cotter G, Davison BA, Felker GM, Filippatos G, Greenberg BH,
has received consulting and/or lecture fees from Bayer, Boehringer Ponikowski P, Unemori E, Voors AA, Adams KF Jr, Dorobantu MI, Grinfeld L,
Jondeau G, Marmor A, Masip J, Pang PS, Werdan K, Prescott MF, Edwards C,
Ingelheim, Roche Diagnostics, Novartis, Orion Pharma, Pfizer, Teichman SL, Trapani A, Bush CA, Saini R, Schumacher C, Severin T, Teerlink
Servier, and Vifor Pharma. The other authors declare no conflicts JR; RELAX-AHF Investigators. Effect of serelaxin on cardiac, renal, and hepatic
of interest. biomarkers in the Relaxin in Acute Heart Failure (RELAX-AHF) development
program: correlation with outcomes. J Am Coll Cardiol 2013;61:196–206.
21. Fuernau G, Poenisch C, Eitel I, Denks D, de Waha S, Poss J, Heine GH, Desch
S, Schuler G, Adams V, Werdan K, Zeymer U, Thiele H. Prognostic impact of
References established and novel renal function biomarkers in myocardial infarction with
1. Rewa O, Bagshaw SM. Acute kidney injury—epidemiology, outcomes and eco- cardiogenic shock: a biomarker substudy of the IABP-SHOCK II-trial. Int J Cardiol
nomics. Nat Rev Nephrol 2014;10:193–207. 2015;191:159–166.

© 2017 The Authors

European Journal of Heart Failure © 2017 European Society of Cardiology
Acute kidney injury in cardiogenic shock
22. Koreny M, Karth GD, Geppert A, Neunteufl T, Priglinger U, Heinz G, Siostrzonek
...............................................
P. Prognosis of patients who develop acute renal failure during the first 24 hours
of cardiogenic shock after myocardial infarction. Am J Med 2002;112:115–119.
23. Marenzi G, Assanelli E, Campodonico J, De Metrio M, Lauri G, Marana I,
Moltrasio M, Rubino M, Veglia F, Montorsi P, Bartorelli AL. Acute kidney injury
in ST-segment elevation acute myocardial infarction complicated by cardiogenic
shock at admission. Crit Care Med 2010;38:438–444.
24. Kellum JA, Sileanu FE, Murugan R, Lucko N, Shaw AD, Clermont G. Classifying
AKI by urine output versus serum creatinine level. J Am Soc Nephrol 2015;
26:2231–2238.
25. Macedo E, Malhotra R, Bouchard J, Wynn SK, Mehta RL. Oliguria is an early
predictor of higher mortality in critically ill patients. Kidney Int 2011;80:760–767.
26. McCullough PA, Bennett-Guerrero E, Chawla LS, Beaver T, Mehta RL, Molitoris
BA, Eldred A, Ball G, Lee HJ, Houser MT, Khan S. ABT-719 for the prevention of
acute kidney injury in patients undergoing high-risk cardiac surgery: a randomized
phase 2b clinical trial. J Am Heart Assoc 2016;5:e003549.
27. Prowle JR, Liu YL, Licari E, Bagshaw SM, Egi M, Haase M, Haase-Fielitz A, Kellum
JA, Cruz D, Ronco C, Tsutsui K, Uchino S, Bellomo R. Oliguria as predictive
biomarker of acute kidney injury in critically ill patients. Crit Care 2011;15:R172.
28. McIlroy DR, Argenziano M, Farkas D, Umann T, Sladen RN. Incorporating
oliguria into the diagnostic criteria for acute kidney injury after on-pump
© 2017 The Authors
European Journal of Heart Failure © 2017 European Society of Cardiology
581
cardiac surgery: impact on incidence and outcomes. J Cardiothorac Vasc Anesth
2013;27:1145–1152.
29. Damman K, van Deursen VM, Navis G, Voors AA, van Veldhuisen DJ, Hillege HL.
Increased central venous pressure is associated with impaired renal function and
mortality in a broad spectrum of patients with cardiovascular disease. J Am Coll
Cardiol 2009;53:582–588.
30. Mullens W, Abrahams Z, Francis GS, Sokos G, Taylor DO, Starling RC, Young JB,
Tang WH. Importance of venous congestion for worsening of renal function in
advanced decompensated heart failure. J Am Coll Cardiol 2009;53:589–596.
31. Hanberg JS, Sury K, Wilson FP, Brisco MA, Ahmad T, Ter Maaten JM, Broughton
JS, Assefa M, Tang WH, Parikh CR, Testani JM. Reduced cardiac index is not the
dominant driver of renal dysfunction in heart failure. J Am Coll Cardiol 2016;67:
2199–2208.
32. Tang WH, Dupont M, Hernandez AF, Voors AA, Hsu AP, Felker GM, Butler J,
Metra M, Anker SD, Troughton RW, Gottlieb SS, McMurray JJ, Armstrong PW,
Massie BM, Califf RM, O’Connor CM, Starling RC. Comparative assessment of
short-term adverse events in acute heart failure with cystatin C and other esti-
mates of renal function: results from the ASCEND-HF trial. JACC Heart Fail 2015;
3:40–49.
33. Odutayo A, Cherney D. Cystatin C and acute changes in glomerular filtration
rate. Clin Nephrol 2012;78:64–75.