BJA Advance Access published December 3, 2014 British Journal of Anaesthesia Page 1 of 9 doi:10.1093/bja/aeu382
Predictive value of urine interleukin-18 in the evolution and outcome of acute kidney injury in critically ill adult patients S. Nisula 1*, R. Yang3, M. Poukkanen 4, S. T. Vaara 1, K. M. Kaukonen1, M. Tallgren 5, M. Haapio 2, J. Tenhunen 6, A. M. Korhonen 1 and V. Pettila¨ 1, The FINNAKI Study Group 1
Intensive Care Units, Division of Anaesthesia and Intensive Care Medicine, Department of Surgery and 2 Department of Medicine, Division of Nephrology, Helsinki University Central Hospital, Box 340, Helsinki 00029, Finland 3 Critical Care Medicine Research Group, Department of Intensive Care Medicine, Tampere University Hospital, Tampere, Finland 4 Department of Anaesthesia and Intensive Care Medicine, Lapland Central Hospital, Rovaniemi, Finland 5 Department of Anesthesia and Intensive Care Medicine, Turku University Hospital, Turku, Finland 6 Department of Surgical Sciences/Anaesthesiology and Intensive Care, University of Uppsala, Uppsala, Sweden * Corresponding author. E-mail: [email protected]
† Acute kidney injury (AKI) is common in the critically ill. † Concentrations of interleukin-18 (IL-18) in urine have been proposed as a new biomarker for AKI. † In this large study, there was only a weak association between urinary IL-18 concentrations within 24 h of ICU admission and the development of AKI. † These data do not support the use of urine IL-18 as a predictor of significant AKI in critically ill adults.
Background. Interleukin-18 (IL-18) is a pro-inflammatory protein, which mediates ischaemic tubular injury, and has been suggested to be a sensitive and specific biomarker for acute kidney injury (AKI). The predictive value of IL-18 in the diagnosis, evolution, and outcome of AKI in critically ill patients is still unclear. Methods. We measured urine IL-18 from critically ill patients at intensive care unit (ICU) admission and 24 h. We evaluated the association of IL-18 with developing new AKI, renal replacement therapy (RRT), and 90-day mortality. We calculated areas under receiver operating characteristics curves (AUCs), best cut-off values, and positive likelihood ratios (LR+) for IL-18 concerning these endpoints. Additionally, we compared the predictive value of IL-18 at ICU admission to that of urine neutrophil gelatinase-associated lipocalin (NGAL). Results. In this study population of 1439 patients the highest urine IL-18 during the first 24 h in the ICU associated with the development of AKI with an AUC [95% confidence interval (CI)] of 0.586 (0.546 –0.627) and with the development of Stage 3 AKI with an AUC (95% CI) of 0.667 (0.591 –0.774). IL-18 predicted the initiation of RRT with an AUC (95% CI) of 0.655 (0.572 – 0.739), and 90-day mortality with an AUC (95% CI) of 0.536 (0.497 –0.574). Conclusions. IL-18 had poor-to-moderate ability to predict AKI, RRT, or 90-day mortality in this large cohort of critically ill patients. Thus, it should be used with caution for diagnostic or predictive purposes in the critically ill. Keywords: acute kidney injury; critical illness; interleukin-18; intensive care; long-term outcome; renal replacement therapy Accepted for publication: 5 August 2014
Acute kidney injury (AKI) is a frequent organ dysfunction in critically ill patients, and it increases both short- and long-term mortality.1 – 4 Current diagnosis and staging of AKI are based on changes in serum creatinine and urine output despite known shortcomings in these variables.5 More specific, sensitive, and rapid biomarkers to identify and monitor AKI are needed.6 Interleukin-18 (IL-18) is a pro-inflammatory protein that acts as an immunoregulatory agent. It has been associated with many autoimmune diseases, ischaemic heart disease, emphysema, metabolic syndrome, and sepsis.7 Regarding AKI, studies first showed that IL-18 mediates tubular injury in mice predisposed to ischaemia,8 and that the lack of IL-18 protects from tubular damage.9 Thereafter, data from humans indicated that urine IL-18 levels are higher in patients with
kidney injury compared either with patients with transient functional renal dysfunction or with healthy controls.10 Based on pathophysiological plausibility urine IL-18 has been suggested as a new biomarker for AKI. However, only few studies have explored the value of urine IL-18 in the diagnosis,11 evolution,12 and outcome of AKI.13 In paediatric patients undergoing cardiac surgery the predictive value of IL-18 to detect AKI has been good with AUCs exceeding 0.8,14 15 but in adult patients,16 and especially critically ill adult patients12 17 the studies have reported discouraging results. In addition, no adequately powered clinical studies exist regarding renal replacement therapy (RRT).18 Studies evaluating IL-18 in the prediction of mortality have been either too small13 or used clinically too short observation periods.11 12
& The Author 2014. Published by Oxford University Press on behalf of the British Journal of Anaesthesia. All rights reserved. For Permissions, please email: [email protected]
Downloaded from http://bja.oxfordjournals.org/ at Laurentian University on December 5, 2014
Editor’s key points
BJA Accordingly, we tested the value of IL-18 separately in prediction of new AKI, RRT, and 90-day mortality in a large cohort of critically ill adult patients. We hypothesized that IL-18 would bring additional benefit to prediction of these outcomes and be useful in clinical practice.
Methods Patients
Definitions We used the KDIGO (kidney disease: improving global outcomes) guidelines5 with both daily creatinine (Cr) and hourly urine output measurements to define AKI. We defined progression of AKI as worsening of KDIGO stage during Day 2 or 3 in the ICU or new onset of AKI on Day 2 or 3. For baseline creatinine we used the latest value from the previous year excluding the week before admission. In patients lacking a baseline Cr, we estimated it by the modification of diet in renal disease (MDRD) equation assuming a glomerular filtration rate of 75 ml min21 1.73 m22 as recommended.19 20 We defined sepsis with the American College of Chest Physicians/ Society of Critical Care Medicine (ACCP/SCCM) guidelines.21 The observation period for AKI and RRT was 3 days from admission because of the known kinetics of IL-18.14 For analyses of AKI and RRT as endpoints, we excluded patients that fulfilled these endpoints already on Day 1.
Data collection To obtain patient characteristics, severity scores, length of stay, and physiological data, we used a prospective study-specific database utilizing the platform of the Finnish Intensive Care Consortium database (Tieto Ltd, Helsinki, Finland). We created a daily case report form (CRF) to record additional patient information such as RRT and sepsis status. The Finnish Population Register Centre provided the 90-day mortality data.
Page 2 of 9
Laboratory samples We collected urine samples from all eligible patients on admission and 24 h later. We aliquoted and stored the samples at 2808C. We used the (Cusabio Biotechw Wuhan, China) ELISA kit for the IL-18 analyses. One author (R.Y., Tampere University Hospital, Tampere, Finland) assayed the samples according to the instructions, and he was blinded to patient data. For IL-18, the measurable range was 3.9 –250 pg litre21. This ELISA method used shows good intra- and inter-assay precision [median coefficient of variation (CV%) ,8% (intra), and ,10% (inter)]. We had urine neutrophil-gelatinase associated lipocalin (NGAL) assayed from a cohort of patients from the FINNAKI study. Urine NGAL was analysed with the (Bioportow NGAL Rapid) ELISA kit. We have previously published NGAL results22 with patient cohorts for IL-18 and NGAL partly (59.4%) overlapping.
Statistical analyses We compared non-parametric data with the Mann –Whitney U-test and categorical variables with the x 2 test or Fisher’s exact test. We present data as medians with interquartile ranges (IQRs) or as absolute numbers [percentage with 95% confidence intervals (CIs)]. We calculated areas under receiver operating characteristics curves (AUCs) with 95% CIs. We defined an AUC of 0.5–0.75 as poor, AUC of 0.75– 0.9 as good, and AUC of .0.9 as excellent as suggested.23 We identified the best cut-off points for IL-18 with the Youden index and calculated sensitivity, specificity, and positive likelihood ratios (LR+), using these cut-off points. We separately tested the following variables: (a) the highest (admission or 24 h) IL-18 concentration (IL-18max), (b) the change in IL-18 concentration from admission to 24 h. Finally we compared (c) admission IL-18 to the admission NGAL, and (d) the combination (IL-18×NGAL) to admission IL-18 and NGAL. We tested the independent predictive value of IL-18max concerning AKI and 90-day mortality by logistic regression analysis. We constructed multivariable models by testing variables in univariable models and selecting significant values (P,0.2) to the multivariable models. Furthermore, we added post hoc analysis regarding extension of observation period to 5 days regarding new AKI and RRT. We performed all analyses with SPSS version 20 and 21 (SPSS, Chicago, IL, USA).
Results Altogether 1439 patients had a urine IL-18 sample available on ICU admission. Of these, 1080 (75.1%) also had an IL-18 sample at 24 h available. The study flow chart is shown in Figure 1. The characteristics of the included study patients are presented in Table 1.
IL-18 and development of new AKI The incidence of AKI in this study population was 497/1439 (34.5%). Of the 497 AKI patients, 229 (46.1%) had AKI on admission day and, thus, we excluded those patients from the analysis regarding new AKI. Of the remaining 1210 patients
Downloaded from http://bja.oxfordjournals.org/ at Laurentian University on December 5, 2014
The Ethics Committee of the Department of Surgery in Helsinki University Hospital gave approval for this study. We collected a written, informed consent from all study patients or proxy. This was a substudy of the prospective, observational, multicentre FINNAKI study.3 The study included consecutive emergency intensive care unit (ICU) admissions and postoperative patients admitted for more than 24 h. The study excluded (i) patients ,18 years of age, (ii) readmitted patients, who received RRT during their previous admission, (iii) patients electively admitted with an ICU length of stay of ,24 h if discharged alive, (iv) patients on chronic dialysis, (v) organ donors, (vi) patients without permanent residency in Finland or without sufficient language skills, (vii) patients transferred between study ICUs if included in the study for 5 days already, and (viii) patients receiving intermediate care. As a result of logistic and economic reasons and to avoid futility we included a convenience sample of patients from the first half of the FINNAKI study (1.9 – 1.12.2011) for this substudy by taking complete sample boxes of consecutive patients out of storage by random blinded for any patient characteristics or studied outcomes.
Nisula et al.
BJA
Interleukin-18 and acute kidney injury
2901 patients
1061 patients after 1.12.2011 401 no urine IL-18 analysed
1439 patients 90-day mortality 20.1%
497 (34.5%) with AKI 229 with AKI on admission day 90-day mortality 31.0%
213 (14.8%) Stage 1 AKI
113 (7.9%) Stage 2 AKI
171 (11.9%) Stage 3 AKI
96 (6.7%) with RRT 47 with RRT on admission day
Fig 1 Flow-chart of the critically ill patients.
268 (22.1%) developed new AKI. 143/1210 (53.4%) patients had new stage 1 AKI, 63/1210 (23.5%) had new stage 2, and 62 (23.1%) had new stage 3 AKI. Figure 2 presents the IL-18 concentrations at admission and at 24 h stratified by presence of AKI in patients with both samples available (n¼1080). The highest IL-18 (IL-18max) predicted new AKI with an AUC (95% CI) of 0.586 (0.546 –0.627), and new stage 3 AKI with an AUC (95% CI) of 0.667 (0.591 – 0.774). IL-18 could predict progression of AKI with an AUC (95% CI) of 0.604 (95% CI 0.563– 0.646). The change in IL-18 from admission to 24 h predicted new AKI with an AUC (95% CI) of 0.557 (0.514 –0.601). In this study cohort chronic kidney disease, Simplified Acute Physiology Score II (SAPS II) (without age and renal points), Sequential Organ Failure Assessment (SOFA) (without renal points), severe sepsis and IL-18 were independently associated with development of AKI. The multivariable model is presented as Table 2.
Interleukin-18 and renal replacement therapy RRT was initiated within 3 days in 96/1439 (6.7%) patients. RRT was initiated on the first ICU day in 47 patients who were excluded from the analyses regarding IL-18 in prediction of RRT. IL-18max predicted RRT with an AUC (95% CI) of 0.655 (0.572 –0.739), and the change in IL-18 with an AUC (95% CI) of 0.531 (0.428 –0.633).
IL-18 and 90-day mortality The crude 90-day mortality was 289/1439 (20.1%, 95% CI 18.0 –22.2%), and the 90-day mortality for patients with AKI was 154/497 (31.0%) compared with 135/942 (14.3%) for patients without AKI. IL-18max predicted 90-day mortality with an AUC (95% CI) of 0.536 (0.497 –0.574), and the change in IL-18 with an AUC (95% CI) of 0.489 (0.447 – 0.532). When tested in a multivariable logistic regression model higher age, liver disease, higher SAPS II score (without
Page 3 of 9
Downloaded from http://bja.oxfordjournals.org/ at Laurentian University on December 5, 2014
942 (65.5%) without AKI 90-day mortality 14.3%
BJA
Nisula et al.
Table 1 Characteristics of critically ill patients (n¼1439) stratified by the presence of AKI. eGFR, estimated glomerular filtration rate (calculated with the MDRD equation); ICU, intensive care unit; SAPS II, Simplified Acute Physiology Score; SOFA, Sequential Organ Failure Assessment Data available from n (%) Age (years)
1439
Gender (male)
1439
Baseline serum/plasma creatinine (mmol litre21)
917 (63.7)
eGFR
917 (63.7)
All patients n51439 n (%) or median [IQR] 63 [50 –73] 920 (63.9) 74 [60 –91] 83.9 [66.3 –104.1]
No AKI n5942 n (%) or median [IQR] 61 [48 –72] 588 (62.4) 73 [59 –88] 86 [69.5 –106.3]
AKI n5497 n (%) or median [IQR] 66 [56 –75] 332 (66.8) 78 [63 –98] 79.4 [60.0 –100.1]
Co-morbidity 1428 (99.2)
122 (8.5)
70 (7.5)
52 (10.6)
Hypertension
1433 (99.6)
664 (46.3)
399 (42.5)
265 (53.5)
Arteriosclerosis
1423(99.9)
170 (11.9)
91 (9.8)
79 (16.0)
Diabetes
1438 (99.9)
326 (22.7)
189 (20.1)
137 (27.6)
Systolic heart failure
1430 (99.9)
165 (11.5)
99 (10.6)
66 (13.4)
Chronic kidney disease
1434 (99.7)
86 (6.0)
35 (3.7)
51 (10.3)
Emergency
1427 (99.2)
1286 (90.1)
832 (89.2)
454 (91.9)
Surgical
1439
485 (33.7)
331 (35.1)
154 (31.0) 65 (13.1)
Admission type
Diagnostic group (APACHE II)
1439
Cardiovascular, operative
182 (12.6)
117 (12.4)
Cardiovascular, non-operative
189 (13.1)
116 (12.3)
73 (14.7)
Respiratory tract, non-operative
178 (12.4)
114 (12.1)
45 (9.1)
Gastrointestinal tract, operative
135 (9.4)
83 (8.8)
52 (10.5)
Metabolic
139 (9.7)
100 (10.6)
39 (7.8)
Neurological, non-operative
133 (9.2)
110 (11.7)
23 (4.6)
Sepsis
89 (6.2)
40 (4.2)
49 (9.9)
Trauma
99 (6.8)
76 (7.9)
23 (4.6)
92 (6.4)
47 (5.0)
45 (9.1)
203 (14.1)
139 (14.8)
64 (12.9)
Gastrointestinal tract, non-operative Other Other clinical variables Highest lactate (day of admission, mmol litre21)
1130 (78.6)
SOFA (highest score, points)
1439
7 [5– 10]
6 [4– 8]
9 [7– 12]
SAPS II score (points)
1439
36 [27 –47]
32 [25 –43]
42 [33 –56]
1.9 [1.2– 3.7]
1.7 [1.1 –2.8]
2.5 [1.5 –4.7]
Mechanical ventilation
1439
961 (66.8)
609 (64.6)
352 (70.8)
Vasoactive treatment
1439
879 (61.1)
499 (53.0)
380 (76.5)
Emergency surgery (,1 week)
1438 (99.9)
329 (22.9)
216 (23.0)
113 (22.7)
Length of ICU stay (days)
1439
Length of hospital stay (days)
1437 (99.9)
age and renal points), acute liver failure, and AKI were independently associated with 90-day mortality in this population (Table 2). The sensitivity, specificity, best cut-off value, and positive likelihood ratio (LR+) with 95% CIs in the prediction of AKI, RRT and 90-day mortality are presented in Table 3. Separate sensitivity analyses excluding (a) patients with sepsis and (b) patients with missing baseline creatinine are included (Table 3).
Interleukin-18 and neutrophil-gelatinase associated lipocalin The admission urine IL-18 and urine NGAL were available for 855 patients (59.4%) of 1439. Comparisons of the predictive
Page 4 of 9
2.6 [1.7– 5.0]
1.9 [1.0 –4.3]
10.00 [5.8– 17.0]
9 [5– 15]
3.7 [2.0 –6.7] 13 [7– 23]
values of IL-18 and NGAL concerning AKI, RRT, and 90-day mortality are presented in Figure 3. Additional post hoc analysis of IL-18 with extension of observation period to 5 days gave the following AUCs (95% CI): 0.586 (0.549 – 0.624) for new AKI, and 0.647 (0.572 – 0.722) for RRT.
Discussion In this study, we showed that urine IL-18 measured at two time points during the first 24 h of ICU admission had only poor-to-moderate ability to predict new AKI, progression of AKI, or initiation of RRT in critically ill adult patients. Also, the association of urine IL-18 with 90-day mortality was weak
Downloaded from http://bja.oxfordjournals.org/ at Laurentian University on December 5, 2014
Chronic obstructive pulmonary disease
BJA
Interleukin-18 and acute kidney injury
No AKI AKI
250.00
IL-18 (pg ml–1)
200.00
150.00
100.00
50.00
Admission
24 h
*P < 0.001 Fig 2 Urine IL-18 concentrations on admission and at 24 h stratified according to whether AKI developed or not during the 3 consecutive days. Included are the patients with IL-18 samples from admission and 24 h available (n¼1080).
and, thus, not clinically useful. Our results strongly challenge the use of IL-18 as a biomarker for these purposes. We report a poor AUC of 0.586 for IL-18 in prediction of AKI developing during the next 48 h. Four reasonably sized11 – 13 17 and one very small study24 have previously reported the association of IL-18 to new AKI in critically ill adult patients, four of these studies with comparably poor AUCs from 0.55 to 0.62.11 12 17 24 The study by Parikh and colleagues,13 which comprised 138 lung injury patients, presented the most encouraging AUC of 0.73 for IL-18 in prediction of AKI. The largest study of IL-18 in an ICU population (comprising 528 patients) presented poor results (AUC 0.55) even though the development of AKI was observed for 7 days.12 A recent meta-analysis also supported our findings: the pooled AUC for IL-18 in prediction of AKI was 0.7 in all populations, and 0.66 in critical care setting.18 In this study the ability of IL-18 to predict new stage 3 AKI instead of any AKI was slightly better (AUC 0.677), but still inadequate for clinical purposes.23 Small studies in cardiac surgery patients have given promise for IL-18 as a AKI progression marker.25 26 However, our study was unable to fortify these results in critically ill patients with an AUC of only 0.604. IL-18 is shown to peak early (0 –24 h) after an insult to the kidneys.12 14 The decreasing IL-18 of AKI patients from admission to 24 h in this study might suggest an earlier peak, and reflect the delay from onset of the illness to ICU admission. The inclusion of cardiac surgery patients, that generally have an isolated insult to the kidneys (cardiopulmonary by-pass), and are admitted to the ICU without delay, could explain the better performance of IL-18 reported in some previous studies.14 27
Page 5 of 9
Downloaded from http://bja.oxfordjournals.org/ at Laurentian University on December 5, 2014
.00
Two previous studies have reported the association between urine IL-18 and RRT.12 17 A multicentre study with a small number of events (n¼19) and an implausible long observation period (19 days)12 reported an AUC of 0.73 for IL-18 in prediction of RRT. Another ICU study with 17 patients fulfilling a composite endpoint of death or dialysis (during 28 days), reported an adjusted OR of 1.76 for urine IL-18.17 Our finding (AUC of 0.655) is in agreement with the literature, and does not support the clinical usefulness of urine IL-18 in prediction of RRT. Only three studies have evaluated urine IL-18 in prediction of mortality in intensive care patients.11 – 13 In our study, we found no association between IL-18 and 90-day mortality supporting the findings by a previous study of 528 critically ill patients demonstrating an AUC of 0.68.12 Some positive results were reported in a study of 339 critically ill patients (AUC 0.83 for 14-day mortality), but the number of patients reaching the mortality endpoint was small (n¼14).11 The ARDS substudy reported an independent association with IL-18 to 28-day mortality (IL-18 .200 pg ml21, hazards ratio 2.32, 95% CI 1.2 –4.4).13 Furthermore, a recent study with serum IL-18 suggested that IL-18 is associated with long-term mortality when measured at RRT initiation.28 In summary, no conclusive data of IL-18 for predicting mortality exist. Data from two previous studies suggest that IL-18 levels are higher in patients with sepsis.17 29 However, our sensitivity analysis excluding patients with sepsis did not significantly change the predictive value of IL-18 (Table 3). Furthermore, we constructed multivariable logistic regression models for AKI and 90-day mortality as outcomes. In the model for AKI, IL-18 was an independent factor but judging from the almost nonaltered model-based AUCs with and without IL-18 and the overall poor AUC of the model (0.697 95% CI 0.653 –0.741 with IL-18) this did not contradict our findings. We found that the admission IL-18 was inferior compared with admission NGAL in prediction of all the endpoints (Fig. 3). In AKI and 90-day mortality as the endpoints both IL-18 and NGAL yielded to poor results in this study. In prediction of RRT NGAL was superior compared with IL-18 with an AUC (0.827) that can be categorized as good. However, RRT initiation as an endpoint is complicated: it lacks unified criteria and the benefit of RRT in various clinical situations is still without conclusive proof.30 Corresponding results concerning comparison of IL-18 and NGAL were previously reported regarding AKI.17 We also found that the combination of IL-18 and NGAL was not better in prediction of AKI, which is in agreement with a previous study comprising 451 patients.17 Our study has some important limitations to consider. First, we did not analyse urine samples from all patients of the FINNAKI study, but chose a convenience sample cohort for this substudy blinded to patients’ characteristics and outcome. Given that our sample size is large and adequate and the patients included did not differ in characteristics from those who were not included we consider our population representative without any plausible selection bias. Secondly, attributable to logistic reasons, we did not centrifuge the urine samples before freezing. However, data show that this does not impact
BJA
Nisula et al.
Table 2 Multivariable logistic regression models for AKI and 90-day mortality including all variables tested in univariable models. Odds ratios (ORs) with 95% CI are given for variables proved significant in multivariable models (P,0.05). 95% CI, 95% confidence interval; AKI, acute kidney injury; corticosteroids (permanent medication); CKD, chronic kidney disease; colloids, starch or gelatin; COPD, chronic obstructive pulmonary disease; IL-18 max, highest IL-18 during first 24 h in the ICU; ns, not significant (P.0.05); NSAID, non-steroid anti-inflammatory drug (permanent medication); OR, odds ratio; RRT, renal replacement therapy; SAPS II, Simplified Acute Physiology Score; SOFA, Sequential Organ Failure Assessment
Age
AKI
90-day mortality
ns
1.045 (1.030 –1.060)
Gender (male)
ns
Diabetes (any) CKD
2.155 (1.029 – 4.510)
ns
COPD
ns
ns
Liver disease
5.305 (2.350 –11.977)
Systolic heart failure
ns ns
ns
Atherosclerosis
ns
ns
NSAID
ns
Corticosteroids
ns
ns
Pre-ICU diuretics
ns
ns
Pre-ICU colloids
ns
Pre-ICU hypotension
ns
ns
Non-operative admission
ns
Emergency surgery (,1 week)
ns
SAPS II (-age minus renal points)
0.974 (0.956 – 0.992)
SOFA (Day 1 min renal points)
1.160 (1.067 – 1.260)
1.060 (1.040 –1.080) ns
Highest lactate (Day 1)
1.103 (1.026 – 1.186)
ns
Severe sepsis
1.687 (1.155 – 2.466)
ns
Acute liver failure
ns
2.748 (1.026 –7.359)
AKI
–
2.064 (1.390 –3.066)
RRT
–
ns
IL-18
1.003 (1.001 – 1.005)
ns
Model-based ROC AUC without IL-18
0.693 (0.649 – 0.737)
0.824 (0.795 –0.854)
Model-based ROC AUC with IL-18
0.697 (0.653 – 0.741)
0.824 (0.795 –0.854)
Table 3 Highest urine IL-18 of 24 h in prediction of AKI, RRT, and 90-day mortality separately. 95% CI, confidence interval; AUC, area under receiver characteristics curve; AKI, acute kidney injury; RRT, renal replacement therapy; LR+, positive likelihood ratio; N/A, not available (numbers inadequate for calculations) Sensitivity
Specificity
AUC (95% CI)
Cut-off value (pg ml21)
LR1
All study patients (n¼1439) New AKI (n¼268)
0.384
0.778
0.586 (0.546 –0.627)
65
1.72 (1.41 – 2.08)
RRT (n¼49)
0.551
0.739
0.655 (0.572 –0.739)
65
2.04 (1.54 – 2.69)
90-day mortality (n¼289)
0.398
0.683
0.536 (0.497 –0.574)
44
1.25 (1.06 – 1.47)
Non-septic patients (n¼749) New AKI (n¼116)
0.319
0.838
0.563 (0.502 –0.624)
65
1.92 (1.38 – 2.67)
RRT (n¼16)
0.688
0.806
0.780 (0.668 –0.893)
65
N/A
90-day mortality (n¼98)
0.306
0.802
0.534 (0.472 –0.597)
64
1.53 (1.10 – 2.14)
Patients with known baseline creatinine (n¼917) New AKI (n¼174)
0.328
0.805
0.570 (0.519 –0.620)
85
1.68 (1.129 –2.20)
RRT (n¼32)
0.563
0.698
0.643 (0.539 –0.748)
54
1.86 (1.35 – 2.57)
90-mortality (n¼212)
0.703
0.524
0.552 (0.508 –0.597)
4
1.11 (1.00 – 1.23)
on the stability of IL-18.31 Thirdly, in some studies the levels of IL-18 were normalized to urine Cr levels, and the recent meta-analysis suggested that normalized values improved the
Page 6 of 9
performance of IL-18 slightly.18 However, this finding is controversial, because the urine creatinine excretion rates are highly variable.32 Thus, we chose to report the results as absolute
Downloaded from http://bja.oxfordjournals.org/ at Laurentian University on December 5, 2014
Hypertension
BJA
Interleukin-18 and acute kidney injury
A
B IL-18 NGAL IL-18*NGAL
1.0
Sensitivity
0.6
0.4
IL-18 NGAL IL-18*NGAL
0.6
0.4
0.0
0.2
0.4
0.6
0.8
1.0
0.0
0.2
0.4
0.6
0.8
1.0
0.0
0.2
RRT
0.6
0.8
1.0
90-day mortality
IL-18
0.531 (0.479–0.584)
IL-18
AUC (95% CI) 0.598 (0.498–0.697)
IL-18
AUC (95% CI) 0.524 (0.476–0.573)
NGAL
0.631 (0.579–0.682)
NGAL
0.827 (0.765–0.889)
NGAL
0.618 (0.573–0.664)
IL-18* NGAL
0.603 (0.553–0.654)
IL-18* NGAL
0.767 (0.708–0.826)
IL-18* NGAL
0.588 (0.542–0.634)
Fig 3 Receiver operating characteristics curves (ROC AUC) (with 95% CI) of admission urine IL-18, admission urine NGAL, and the combination of urine IL-18 and NGAL in the prediction of AKI, RRT, and 90-day mortality in critically ill patients (n¼855).
concentrations without normalization to creatinine. Fourthly, we had the baseline creatinine available from only 64% patients, and we were obliged to use the recommended19 MDRD equation (with its limitations) to estimate the baseline value if missing. According to the literature, this may lead to over- or underestimation of AKI.33 – 35 Furthermore, to decrease any bias induced by the use of MDRD, we performed a sensitivity analysis excluding patients lacking the baseline creatinine, which procedure did not affect our results. Thus, we consider our findings to be reliable and also generalizable because of the large random sample from consecutive critically ill patients. An obvious strength of our study is that it was a large, prospective, multicentre study evaluating the usefulness of urine IL-18 in the prediction of multiple clinically important patientcentred outcomes (AKI, RRT, 90-day mortality) in critical care. Thus, we consider our findings to be reliable and also generalizable. Moreover, we aimed to construct a plausible design to assess the predictive properties of this biomarker using admission and 24 h samples, a reasonably short 3-day observation period for AKI and RRT with exclusion of patients with AKI or RRT already on admission. Finally, our AKI classification used the latest KDIGO criteria5 including both daily creatinine and hourly urine output measurements. To conclude, shown by the only poor-to-moderate AUCs and fortified by the unaltered predictive model-based AUCs IL-18 has inadequate properties to predict AKI, RRT, or 90-day mortality in critically ill patients.
Authors’ contributions S.N.: drafted the manuscript, performed the statistical analyses, and participated in the design and data gathering of the study. R.Y.: performed the laboratory analyses and revised the
manuscript. M.P.: participated in the data gathering of the study and revised the manuscript. S.T.V.: participated in the design and data gathering of the study and helped to draft the manuscript and to perform the statistical analyses. K.M.K.: participated in the design and coordination of the study, and revised the manuscript. M.T.: participated in the data gathering of the study and revised the manuscript. M.H.: participated in the design and data gathering of the study and revised the manuscript. J.T.: participated in the design of the study and coordination of the laboratory analyses and revised the manuscript. A.M.K.: participated in the design and co-ordination of the study, collected study data, and helped to draft the manuscript. V.P.: principal investigator of the study and participated in the design and coordination of the study, and helped to draft the manuscript and to perform the statistical analyses.
Acknowledgements We are grateful for all the members of the FINNAKI study group in participating hospitals and Tieto Healthcare and Welfare Ltd for database management.
Declaration of interest S.N., R.Y., M.P., S.T.V., K.M.K., M.T., M.H., J.T., A.M.K., and V.P. declare no conflict of interest.
Funding Clinical Research funding (E.V.O.) TYH 2010109/2011210 and T102010070 from Helsinki University Hospital (V.P.), and grants from the Finnish Society of Intensive Care, the Academy of Finland (K.M.K.), the Juselius Foundation (V.P.) and the Pa¨ivikki and Sakari Sohlberg Foundation (V.P.), and the Finnish Society of Anaesthesiologists (S.N.). Laboratory
Page 7 of 9
Downloaded from http://bja.oxfordjournals.org/ at Laurentian University on December 5, 2014
AUC (95% CI)
0.4
1–Specificity
1–Specificity
1–Specificity AKI
0.4
0.0
0.0
0.0
0.6
0.2
0.2
0.2
IL-18 NGAL IL-18*NGAL
0.8
0.8
Sensitivity
0.8
Sensitivity
C 1.0
1.0
BJA analyses were funded by a grant from the Academy of Finland (K.M.K.) and external funding for Critical Care Medicine Research Group (J.T.).
References
Page 8 of 9
17
18
19 20
21
22
23 24
25
26
27
28
29
30 31 32
33
34
35
biomarkers of nephron damage: a multicenter prospective cohort study. J Am Coll Cardiol 2012; 59: 246–55 Siew ED, Ikizler TA, Gebretsadik T, et al. Elevated urinary IL-18 levels at the time of ICU admission predict adverse clinical outcomes. Clin J Am Soc Nephrol 2010; 5: 1497– 505 Liu Y, Guo W, Zhang J, et al. Urinary interleukin 18 for detection of acute kidney injury: a meta-analysis. Am J Kidney Dis 2013; 62: 1058– 67 Kellum JA, Bellomo R, Ronco C. Acute Dialysis Quality Initiative (ADQI): methodology. Int J Artif Organs 2008; 31: 90– 3 Levey AS, Coresh J, Balk E, et al. National Kidney Foundation practice guidelines for chronic kidney disease: evaluation, classification, and stratification. Ann Intern Med 2003; 139: 137–47 Levy MM, Fink MP, Marshall JC, et al. 2001 SCCM/ESICM/ACCP/ATS/ SIS International Sepsis Definitions Conference. Intensive Care Med 2003; 29: 530– 8 Nisula S, Yang R, Kaukonen KM, et al. The urine protein NGAL predicts renal replacement therapy, but not acute kidney injury or 90-day mortality in critically ill adult patients. Anesth Analg 2014; 119: 95– 102 Ray P, Le Manach Y, Riou B, Houle TT. Statistical evaluation of a biomarker. Anesthesiology 2010; 112: 1023– 40 Metzger J, Kirsch T, Schiffer E, et al. Urinary excretion of twenty peptides forms an early and accurate diagnostic pattern of acute kidney injury. Kidney Int 2010; 78: 1252–62 Arthur JM, Hill EG, Alge JL, et al. Evaluation of 32 urine biomarkers to predict the progression of acute kidney injury after cardiac surgery. Kidney Int 2014; 85: 431–8 Koyner JL, Garg AX, Coca SG, et al. Biomarkers predict progression of acute kidney injury after cardiac surgery. J Am Soc Nephrol 2012; 23: 905– 14 Parikh CR, Mishra J, Thiessen-Philbrook H, et al. Urinary IL-18 is an early predictive biomarker of acute kidney injury after cardiac surgery. Kidney Int 2006; 70: 199–203 Lin CY, Chang CH, Fan PC, et al. Serum interleukin-18 at commencement of renal replacement therapy predicts short-term prognosis in critically ill patients with acute kidney injury. PloS One 2013; 8: e66028 Washburn KK, Zappitelli M, Arikan AA, et al. Urinary interleukin-18 is an acute kidney injury biomarker in critically ill children. Nephrol Dial Transplant 2008; 23: 566–72 Joannidis M, Forni LG. Clinical review: timing of renal replacement therapy. Crit Care 2011; 15: 223 Parikh CR, Butrymowicz I, Yu A, et al. Urine stability studies for novel biomarkers of acute kidney injury. Am J Kidney Dis 2013; 63: 567– 72 Waikar SS, Sabbisetti VS, Bonventre JV. Normalization of urinary biomarkers to creatinine during changes in glomerular filtration rate. Kidney Int 2010; 78: 486–94 Bagshaw SM, Uchino S, Cruz D, et al. A comparison of observed versus estimated baseline creatinine for determination of RIFLE class in patients with acute kidney injury. Nephrol Dial Transplant 2009; 24: 2739– 44 Gaiao S, Cruz DN. Baseline creatinine to define acute kidney injury: is there any consensus? Nephrol Dial Transplant 2010; 25: 3812 – 4 Poggio ED, Nef PC, Wang X, et al. Performance of the CockcroftGault and modification of diet in renal disease equations in estimating GFR in ill hospitalized patients. Am J Kidney Dis 2005; 46: 242– 52
Downloaded from http://bja.oxfordjournals.org/ at Laurentian University on December 5, 2014
1 Bagshaw SM, George C, Dinu I, Bellomo R. A multi-centre evaluation of the RIFLE criteria for early acute kidney injury in critically ill patients. Nephrol Dial Transplant 2008; 23: 1203–10 2 Joannidis M, Metnitz B, Bauer P, et al. Acute kidney injury in critically ill patients classified by AKIN versus RIFLE using the SAPS 3 database. Intensive Care Med 2009; 35: 1692– 702 3 Nisula S, Kaukonen KM, Vaara ST, et al. Incidence, risk factors and 90-day mortality of patients with acute kidney injury in Finnish intensive care units: the FINNAKI study. Intensive Care Med 2013; 39: 420–8 4 Vaara ST, Pettila¨ V, Kaukonen KM, et al. The attributable mortality of acute kidney injury: a sequentially matched analysis. Crit Care Med 2013; 42: 878– 85 5 Kidney Disease: Improving Global Outcomes (KDIGO) Acute Kidney Injury Work Group. KDIGO Clinical Practice Guideline for Acute Kidney Injury. Kidney Inter, Suppl 2012; 2: 1– 138 6 McCullough PA, Shaw AD, Haase M, et al. Diagnosis of acute kidney injury using functional and injury biomarkers: workgroup statements from the tenth Acute Dialysis Quality Initiative Consensus Conference. Contrib Nephrol 2013; 182: 13 –29 7 Dinarello CA, Novick D, Kim S, Kaplanski G. Interleukin-18 and IL-18 binding protein. Front Immunol 2013; 4: 289 8 Melnikov VY, Faubel S, Siegmund B, Lucia MS, Ljubanovic D, Edelstein CL. Neutrophil-independent mechanisms of caspase-1and IL-18-mediated ischemic acute tubular necrosis in mice. J Clin Invest 2002; 110: 1083–91 9 Melnikov VY, Ecder T, Fantuzzi G, et al. Impaired IL-18 processing protects caspase-1-deficient mice from ischemic acute renal failure. J Clin Invest 2001; 107: 1145– 52 10 Parikh CR, Jani A, Melnikov VY, Faubel S, Edelstein CL. Urinary interleukin-18 is a marker of human acute tubular necrosis. Am J Kidney Dis 2004; 43: 405–14 11 Doi K, Negishi K, Ishizu T, et al. Evaluation of new acute kidney injury biomarkers in a mixed intensive care unit. Crit Care Med 2011; 39: 2464– 9 12 Endre ZH, Pickering JW, Walker RJ, et al. Improved performance of urinary biomarkers of acute kidney injury in the critically ill by stratification for injury duration and baseline renal function. Kidney Int 2011; 79: 1119– 30 13 Parikh CR, Abraham E, Ancukiewicz M, Edelstein CL. Urine IL-18 is an early diagnostic marker for acute kidney injury and predicts mortality in the intensive care unit. Clin J Am Soc Nephrol 2005; 16: 3046– 52 14 Krawczeski CD, Goldstein SL, Woo JG, et al. Temporal relationship and predictive value of urinary acute kidney injury biomarkers after pediatric cardiopulmonary bypass. J Am Coll Cardiol 2011; 58: 2301–9 15 Zheng J, Xiao Y, Yao Y, et al. Comparison of urinary biomarkers for early detection of acute kidney injury after cardiopulmonary bypass surgery in infants and young children. Pediatr Cardiol 2013; 34: 880–6 16 Nickolas TL, Schmidt-Ott KM, Canetta P, et al. Diagnostic and prognostic stratification in the emergency department using urinary
Nisula et al.
Interleukin-18 and acute kidney injury
Appendix
Central Hospital: Meri Poukkanen, Esa Lintula, Sirpa Suominen. La¨nsi Pohja Central Hospital: Jorma Heikkinen, Timo Lavander, Kirsi Heinonen, Anne-Mari Juopperi. Middle Ostrobothnia Central Hospital: Tadeusz Kaminski, Fiia Ga¨ddna¨s, Tuija Kuusela, Jane Roiko. North Karelia Central Hospital: Sari Karlsson, Matti Reinikainen, Tero Surakka, Helena Jyrko¨nen, Tanja Eiserbeck, Jaana Kallinen. Satakunta Hospital district: Vesa Lund, Pa¨ivi Tuominen, Pauliina Perkola, Riikka Tuominen, Marika Hietaranta, Satu Johansson. South Karelia Central Hospital: Seppo Hovilehto, Anne Kirsi, Pekka Tiainen, Tuija Mylla¨rinen, Pirjo Leino, Anne Toropainen. Tampere University Hospital: Anne Kuitunen, Ilona Leppa¨nen, Markus Levoranta, Sanna Hoppu, Jukka Sauranen, Jyrki Tenhunen, Atte Kukkurainen, Samuli Kortelainen, Simo Varila. Turku University Hospital: Outi Inkinen, Niina Koivuviita, Jutta Kotama¨ki, Anu Laine. Oulu University Hospital: Tero Ala-Kokko, Jouko Laurila, Sinikka Sa¨lkio¨. Vaasa Central Hospital: Simo-Pekka Koivisto, Raku Hautama¨ki, Maria Skinnar. Handling editor: J. P. Thompson
Page 9 of 9
Downloaded from http://bja.oxfordjournals.org/ at Laurentian University on December 5, 2014
The FINNAKI study Group Central Finland Central Hospital: Raili Laru-Sompa, Anni Pulkkinen, Minna Saarelainen, Mikko Reilama, Sinikka Tolmunen, Ulla Rantalainen, Marja Miettinen. East Savo Central Hospital: Markku Suvela, Katrine Pesola, Pekka Saastamoinen, Sirpa Kauppinen. Helsinki University Central Hospital: Ville Pettila¨, Kirsi-Maija Kaukonen, Anna-Maija Korhonen, Sara Nisula, Suvi Vaara, Raili Suojaranta-Ylinen, Leena Mildh, Mikko Haapio, Laura Nurminen, Sari Sutinen, Leena Pettila¨, Helina¨ Laitinen, Heidi Syrja¨, Kirsi Henttonen, Elina Lappi, Hillevi Boman. Jorvi Central Hospital: Tero Varpula, Pa¨ivi Porkka, Mirka Sivula Mira Rahkonen, Anne Tsurkka, Taina Nieminen, Niina Prittinen. Kanta-Ha¨me Central hospital: Ari Alaspa¨a¨, Ville Salanto, Hanna Juntunen, Teija Sanisalo. Kuopio University Hospital: Ilkka Parviainen, Ari Uusaro, Esko Ruokonen, Stepani Bendel, Niina Rissanen, Maarit La˚ng, Sari Rahikainen, Saija Rissanen, Merja Ahonen, Elina Halonen, Eija Vaskelainen. Lapland
BJA
Predictive value of urine interleukin-18 in the evolution and outcome of acute kidney injury in critically ill adult patients S. Nisula 1*, R. Yang3, M. Poukkanen 4, S. T. Vaara 1, K. M. Kaukonen1, M. Tallgren 5, M. Haapio 2, J. Tenhunen 6, A. M. Korhonen 1 and V. Pettila¨ 1, The FINNAKI Study Group 1
Intensive Care Units, Division of Anaesthesia and Intensive Care Medicine, Department of Surgery and 2 Department of Medicine, Division of Nephrology, Helsinki University Central Hospital, Box 340, Helsinki 00029, Finland 3 Critical Care Medicine Research Group, Department of Intensive Care Medicine, Tampere University Hospital, Tampere, Finland 4 Department of Anaesthesia and Intensive Care Medicine, Lapland Central Hospital, Rovaniemi, Finland 5 Department of Anesthesia and Intensive Care Medicine, Turku University Hospital, Turku, Finland 6 Department of Surgical Sciences/Anaesthesiology and Intensive Care, University of Uppsala, Uppsala, Sweden * Corresponding author. E-mail: [email protected]
† Acute kidney injury (AKI) is common in the critically ill. † Concentrations of interleukin-18 (IL-18) in urine have been proposed as a new biomarker for AKI. † In this large study, there was only a weak association between urinary IL-18 concentrations within 24 h of ICU admission and the development of AKI. † These data do not support the use of urine IL-18 as a predictor of significant AKI in critically ill adults.
Background. Interleukin-18 (IL-18) is a pro-inflammatory protein, which mediates ischaemic tubular injury, and has been suggested to be a sensitive and specific biomarker for acute kidney injury (AKI). The predictive value of IL-18 in the diagnosis, evolution, and outcome of AKI in critically ill patients is still unclear. Methods. We measured urine IL-18 from critically ill patients at intensive care unit (ICU) admission and 24 h. We evaluated the association of IL-18 with developing new AKI, renal replacement therapy (RRT), and 90-day mortality. We calculated areas under receiver operating characteristics curves (AUCs), best cut-off values, and positive likelihood ratios (LR+) for IL-18 concerning these endpoints. Additionally, we compared the predictive value of IL-18 at ICU admission to that of urine neutrophil gelatinase-associated lipocalin (NGAL). Results. In this study population of 1439 patients the highest urine IL-18 during the first 24 h in the ICU associated with the development of AKI with an AUC [95% confidence interval (CI)] of 0.586 (0.546 –0.627) and with the development of Stage 3 AKI with an AUC (95% CI) of 0.667 (0.591 –0.774). IL-18 predicted the initiation of RRT with an AUC (95% CI) of 0.655 (0.572 – 0.739), and 90-day mortality with an AUC (95% CI) of 0.536 (0.497 –0.574). Conclusions. IL-18 had poor-to-moderate ability to predict AKI, RRT, or 90-day mortality in this large cohort of critically ill patients. Thus, it should be used with caution for diagnostic or predictive purposes in the critically ill. Keywords: acute kidney injury; critical illness; interleukin-18; intensive care; long-term outcome; renal replacement therapy Accepted for publication: 5 August 2014
Acute kidney injury (AKI) is a frequent organ dysfunction in critically ill patients, and it increases both short- and long-term mortality.1 – 4 Current diagnosis and staging of AKI are based on changes in serum creatinine and urine output despite known shortcomings in these variables.5 More specific, sensitive, and rapid biomarkers to identify and monitor AKI are needed.6 Interleukin-18 (IL-18) is a pro-inflammatory protein that acts as an immunoregulatory agent. It has been associated with many autoimmune diseases, ischaemic heart disease, emphysema, metabolic syndrome, and sepsis.7 Regarding AKI, studies first showed that IL-18 mediates tubular injury in mice predisposed to ischaemia,8 and that the lack of IL-18 protects from tubular damage.9 Thereafter, data from humans indicated that urine IL-18 levels are higher in patients with
kidney injury compared either with patients with transient functional renal dysfunction or with healthy controls.10 Based on pathophysiological plausibility urine IL-18 has been suggested as a new biomarker for AKI. However, only few studies have explored the value of urine IL-18 in the diagnosis,11 evolution,12 and outcome of AKI.13 In paediatric patients undergoing cardiac surgery the predictive value of IL-18 to detect AKI has been good with AUCs exceeding 0.8,14 15 but in adult patients,16 and especially critically ill adult patients12 17 the studies have reported discouraging results. In addition, no adequately powered clinical studies exist regarding renal replacement therapy (RRT).18 Studies evaluating IL-18 in the prediction of mortality have been either too small13 or used clinically too short observation periods.11 12
& The Author 2014. Published by Oxford University Press on behalf of the British Journal of Anaesthesia. All rights reserved. For Permissions, please email: [email protected]
Downloaded from http://bja.oxfordjournals.org/ at Laurentian University on December 5, 2014
Editor’s key points
BJA Accordingly, we tested the value of IL-18 separately in prediction of new AKI, RRT, and 90-day mortality in a large cohort of critically ill adult patients. We hypothesized that IL-18 would bring additional benefit to prediction of these outcomes and be useful in clinical practice.
Methods Patients
Definitions We used the KDIGO (kidney disease: improving global outcomes) guidelines5 with both daily creatinine (Cr) and hourly urine output measurements to define AKI. We defined progression of AKI as worsening of KDIGO stage during Day 2 or 3 in the ICU or new onset of AKI on Day 2 or 3. For baseline creatinine we used the latest value from the previous year excluding the week before admission. In patients lacking a baseline Cr, we estimated it by the modification of diet in renal disease (MDRD) equation assuming a glomerular filtration rate of 75 ml min21 1.73 m22 as recommended.19 20 We defined sepsis with the American College of Chest Physicians/ Society of Critical Care Medicine (ACCP/SCCM) guidelines.21 The observation period for AKI and RRT was 3 days from admission because of the known kinetics of IL-18.14 For analyses of AKI and RRT as endpoints, we excluded patients that fulfilled these endpoints already on Day 1.
Data collection To obtain patient characteristics, severity scores, length of stay, and physiological data, we used a prospective study-specific database utilizing the platform of the Finnish Intensive Care Consortium database (Tieto Ltd, Helsinki, Finland). We created a daily case report form (CRF) to record additional patient information such as RRT and sepsis status. The Finnish Population Register Centre provided the 90-day mortality data.
Page 2 of 9
Laboratory samples We collected urine samples from all eligible patients on admission and 24 h later. We aliquoted and stored the samples at 2808C. We used the (Cusabio Biotechw Wuhan, China) ELISA kit for the IL-18 analyses. One author (R.Y., Tampere University Hospital, Tampere, Finland) assayed the samples according to the instructions, and he was blinded to patient data. For IL-18, the measurable range was 3.9 –250 pg litre21. This ELISA method used shows good intra- and inter-assay precision [median coefficient of variation (CV%) ,8% (intra), and ,10% (inter)]. We had urine neutrophil-gelatinase associated lipocalin (NGAL) assayed from a cohort of patients from the FINNAKI study. Urine NGAL was analysed with the (Bioportow NGAL Rapid) ELISA kit. We have previously published NGAL results22 with patient cohorts for IL-18 and NGAL partly (59.4%) overlapping.
Statistical analyses We compared non-parametric data with the Mann –Whitney U-test and categorical variables with the x 2 test or Fisher’s exact test. We present data as medians with interquartile ranges (IQRs) or as absolute numbers [percentage with 95% confidence intervals (CIs)]. We calculated areas under receiver operating characteristics curves (AUCs) with 95% CIs. We defined an AUC of 0.5–0.75 as poor, AUC of 0.75– 0.9 as good, and AUC of .0.9 as excellent as suggested.23 We identified the best cut-off points for IL-18 with the Youden index and calculated sensitivity, specificity, and positive likelihood ratios (LR+), using these cut-off points. We separately tested the following variables: (a) the highest (admission or 24 h) IL-18 concentration (IL-18max), (b) the change in IL-18 concentration from admission to 24 h. Finally we compared (c) admission IL-18 to the admission NGAL, and (d) the combination (IL-18×NGAL) to admission IL-18 and NGAL. We tested the independent predictive value of IL-18max concerning AKI and 90-day mortality by logistic regression analysis. We constructed multivariable models by testing variables in univariable models and selecting significant values (P,0.2) to the multivariable models. Furthermore, we added post hoc analysis regarding extension of observation period to 5 days regarding new AKI and RRT. We performed all analyses with SPSS version 20 and 21 (SPSS, Chicago, IL, USA).
Results Altogether 1439 patients had a urine IL-18 sample available on ICU admission. Of these, 1080 (75.1%) also had an IL-18 sample at 24 h available. The study flow chart is shown in Figure 1. The characteristics of the included study patients are presented in Table 1.
IL-18 and development of new AKI The incidence of AKI in this study population was 497/1439 (34.5%). Of the 497 AKI patients, 229 (46.1%) had AKI on admission day and, thus, we excluded those patients from the analysis regarding new AKI. Of the remaining 1210 patients
Downloaded from http://bja.oxfordjournals.org/ at Laurentian University on December 5, 2014
The Ethics Committee of the Department of Surgery in Helsinki University Hospital gave approval for this study. We collected a written, informed consent from all study patients or proxy. This was a substudy of the prospective, observational, multicentre FINNAKI study.3 The study included consecutive emergency intensive care unit (ICU) admissions and postoperative patients admitted for more than 24 h. The study excluded (i) patients ,18 years of age, (ii) readmitted patients, who received RRT during their previous admission, (iii) patients electively admitted with an ICU length of stay of ,24 h if discharged alive, (iv) patients on chronic dialysis, (v) organ donors, (vi) patients without permanent residency in Finland or without sufficient language skills, (vii) patients transferred between study ICUs if included in the study for 5 days already, and (viii) patients receiving intermediate care. As a result of logistic and economic reasons and to avoid futility we included a convenience sample of patients from the first half of the FINNAKI study (1.9 – 1.12.2011) for this substudy by taking complete sample boxes of consecutive patients out of storage by random blinded for any patient characteristics or studied outcomes.
Nisula et al.
BJA
Interleukin-18 and acute kidney injury
2901 patients
1061 patients after 1.12.2011 401 no urine IL-18 analysed
1439 patients 90-day mortality 20.1%
497 (34.5%) with AKI 229 with AKI on admission day 90-day mortality 31.0%
213 (14.8%) Stage 1 AKI
113 (7.9%) Stage 2 AKI
171 (11.9%) Stage 3 AKI
96 (6.7%) with RRT 47 with RRT on admission day
Fig 1 Flow-chart of the critically ill patients.
268 (22.1%) developed new AKI. 143/1210 (53.4%) patients had new stage 1 AKI, 63/1210 (23.5%) had new stage 2, and 62 (23.1%) had new stage 3 AKI. Figure 2 presents the IL-18 concentrations at admission and at 24 h stratified by presence of AKI in patients with both samples available (n¼1080). The highest IL-18 (IL-18max) predicted new AKI with an AUC (95% CI) of 0.586 (0.546 –0.627), and new stage 3 AKI with an AUC (95% CI) of 0.667 (0.591 – 0.774). IL-18 could predict progression of AKI with an AUC (95% CI) of 0.604 (95% CI 0.563– 0.646). The change in IL-18 from admission to 24 h predicted new AKI with an AUC (95% CI) of 0.557 (0.514 –0.601). In this study cohort chronic kidney disease, Simplified Acute Physiology Score II (SAPS II) (without age and renal points), Sequential Organ Failure Assessment (SOFA) (without renal points), severe sepsis and IL-18 were independently associated with development of AKI. The multivariable model is presented as Table 2.
Interleukin-18 and renal replacement therapy RRT was initiated within 3 days in 96/1439 (6.7%) patients. RRT was initiated on the first ICU day in 47 patients who were excluded from the analyses regarding IL-18 in prediction of RRT. IL-18max predicted RRT with an AUC (95% CI) of 0.655 (0.572 –0.739), and the change in IL-18 with an AUC (95% CI) of 0.531 (0.428 –0.633).
IL-18 and 90-day mortality The crude 90-day mortality was 289/1439 (20.1%, 95% CI 18.0 –22.2%), and the 90-day mortality for patients with AKI was 154/497 (31.0%) compared with 135/942 (14.3%) for patients without AKI. IL-18max predicted 90-day mortality with an AUC (95% CI) of 0.536 (0.497 –0.574), and the change in IL-18 with an AUC (95% CI) of 0.489 (0.447 – 0.532). When tested in a multivariable logistic regression model higher age, liver disease, higher SAPS II score (without
Page 3 of 9
Downloaded from http://bja.oxfordjournals.org/ at Laurentian University on December 5, 2014
942 (65.5%) without AKI 90-day mortality 14.3%
BJA
Nisula et al.
Table 1 Characteristics of critically ill patients (n¼1439) stratified by the presence of AKI. eGFR, estimated glomerular filtration rate (calculated with the MDRD equation); ICU, intensive care unit; SAPS II, Simplified Acute Physiology Score; SOFA, Sequential Organ Failure Assessment Data available from n (%) Age (years)
1439
Gender (male)
1439
Baseline serum/plasma creatinine (mmol litre21)
917 (63.7)
eGFR
917 (63.7)
All patients n51439 n (%) or median [IQR] 63 [50 –73] 920 (63.9) 74 [60 –91] 83.9 [66.3 –104.1]
No AKI n5942 n (%) or median [IQR] 61 [48 –72] 588 (62.4) 73 [59 –88] 86 [69.5 –106.3]
AKI n5497 n (%) or median [IQR] 66 [56 –75] 332 (66.8) 78 [63 –98] 79.4 [60.0 –100.1]
Co-morbidity 1428 (99.2)
122 (8.5)
70 (7.5)
52 (10.6)
Hypertension
1433 (99.6)
664 (46.3)
399 (42.5)
265 (53.5)
Arteriosclerosis
1423(99.9)
170 (11.9)
91 (9.8)
79 (16.0)
Diabetes
1438 (99.9)
326 (22.7)
189 (20.1)
137 (27.6)
Systolic heart failure
1430 (99.9)
165 (11.5)
99 (10.6)
66 (13.4)
Chronic kidney disease
1434 (99.7)
86 (6.0)
35 (3.7)
51 (10.3)
Emergency
1427 (99.2)
1286 (90.1)
832 (89.2)
454 (91.9)
Surgical
1439
485 (33.7)
331 (35.1)
154 (31.0) 65 (13.1)
Admission type
Diagnostic group (APACHE II)
1439
Cardiovascular, operative
182 (12.6)
117 (12.4)
Cardiovascular, non-operative
189 (13.1)
116 (12.3)
73 (14.7)
Respiratory tract, non-operative
178 (12.4)
114 (12.1)
45 (9.1)
Gastrointestinal tract, operative
135 (9.4)
83 (8.8)
52 (10.5)
Metabolic
139 (9.7)
100 (10.6)
39 (7.8)
Neurological, non-operative
133 (9.2)
110 (11.7)
23 (4.6)
Sepsis
89 (6.2)
40 (4.2)
49 (9.9)
Trauma
99 (6.8)
76 (7.9)
23 (4.6)
92 (6.4)
47 (5.0)
45 (9.1)
203 (14.1)
139 (14.8)
64 (12.9)
Gastrointestinal tract, non-operative Other Other clinical variables Highest lactate (day of admission, mmol litre21)
1130 (78.6)
SOFA (highest score, points)
1439
7 [5– 10]
6 [4– 8]
9 [7– 12]
SAPS II score (points)
1439
36 [27 –47]
32 [25 –43]
42 [33 –56]
1.9 [1.2– 3.7]
1.7 [1.1 –2.8]
2.5 [1.5 –4.7]
Mechanical ventilation
1439
961 (66.8)
609 (64.6)
352 (70.8)
Vasoactive treatment
1439
879 (61.1)
499 (53.0)
380 (76.5)
Emergency surgery (,1 week)
1438 (99.9)
329 (22.9)
216 (23.0)
113 (22.7)
Length of ICU stay (days)
1439
Length of hospital stay (days)
1437 (99.9)
age and renal points), acute liver failure, and AKI were independently associated with 90-day mortality in this population (Table 2). The sensitivity, specificity, best cut-off value, and positive likelihood ratio (LR+) with 95% CIs in the prediction of AKI, RRT and 90-day mortality are presented in Table 3. Separate sensitivity analyses excluding (a) patients with sepsis and (b) patients with missing baseline creatinine are included (Table 3).
Interleukin-18 and neutrophil-gelatinase associated lipocalin The admission urine IL-18 and urine NGAL were available for 855 patients (59.4%) of 1439. Comparisons of the predictive
Page 4 of 9
2.6 [1.7– 5.0]
1.9 [1.0 –4.3]
10.00 [5.8– 17.0]
9 [5– 15]
3.7 [2.0 –6.7] 13 [7– 23]
values of IL-18 and NGAL concerning AKI, RRT, and 90-day mortality are presented in Figure 3. Additional post hoc analysis of IL-18 with extension of observation period to 5 days gave the following AUCs (95% CI): 0.586 (0.549 – 0.624) for new AKI, and 0.647 (0.572 – 0.722) for RRT.
Discussion In this study, we showed that urine IL-18 measured at two time points during the first 24 h of ICU admission had only poor-to-moderate ability to predict new AKI, progression of AKI, or initiation of RRT in critically ill adult patients. Also, the association of urine IL-18 with 90-day mortality was weak
Downloaded from http://bja.oxfordjournals.org/ at Laurentian University on December 5, 2014
Chronic obstructive pulmonary disease
BJA
Interleukin-18 and acute kidney injury
No AKI AKI
250.00
IL-18 (pg ml–1)
200.00
150.00
100.00
50.00
Admission
24 h
*P < 0.001 Fig 2 Urine IL-18 concentrations on admission and at 24 h stratified according to whether AKI developed or not during the 3 consecutive days. Included are the patients with IL-18 samples from admission and 24 h available (n¼1080).
and, thus, not clinically useful. Our results strongly challenge the use of IL-18 as a biomarker for these purposes. We report a poor AUC of 0.586 for IL-18 in prediction of AKI developing during the next 48 h. Four reasonably sized11 – 13 17 and one very small study24 have previously reported the association of IL-18 to new AKI in critically ill adult patients, four of these studies with comparably poor AUCs from 0.55 to 0.62.11 12 17 24 The study by Parikh and colleagues,13 which comprised 138 lung injury patients, presented the most encouraging AUC of 0.73 for IL-18 in prediction of AKI. The largest study of IL-18 in an ICU population (comprising 528 patients) presented poor results (AUC 0.55) even though the development of AKI was observed for 7 days.12 A recent meta-analysis also supported our findings: the pooled AUC for IL-18 in prediction of AKI was 0.7 in all populations, and 0.66 in critical care setting.18 In this study the ability of IL-18 to predict new stage 3 AKI instead of any AKI was slightly better (AUC 0.677), but still inadequate for clinical purposes.23 Small studies in cardiac surgery patients have given promise for IL-18 as a AKI progression marker.25 26 However, our study was unable to fortify these results in critically ill patients with an AUC of only 0.604. IL-18 is shown to peak early (0 –24 h) after an insult to the kidneys.12 14 The decreasing IL-18 of AKI patients from admission to 24 h in this study might suggest an earlier peak, and reflect the delay from onset of the illness to ICU admission. The inclusion of cardiac surgery patients, that generally have an isolated insult to the kidneys (cardiopulmonary by-pass), and are admitted to the ICU without delay, could explain the better performance of IL-18 reported in some previous studies.14 27
Page 5 of 9
Downloaded from http://bja.oxfordjournals.org/ at Laurentian University on December 5, 2014
.00
Two previous studies have reported the association between urine IL-18 and RRT.12 17 A multicentre study with a small number of events (n¼19) and an implausible long observation period (19 days)12 reported an AUC of 0.73 for IL-18 in prediction of RRT. Another ICU study with 17 patients fulfilling a composite endpoint of death or dialysis (during 28 days), reported an adjusted OR of 1.76 for urine IL-18.17 Our finding (AUC of 0.655) is in agreement with the literature, and does not support the clinical usefulness of urine IL-18 in prediction of RRT. Only three studies have evaluated urine IL-18 in prediction of mortality in intensive care patients.11 – 13 In our study, we found no association between IL-18 and 90-day mortality supporting the findings by a previous study of 528 critically ill patients demonstrating an AUC of 0.68.12 Some positive results were reported in a study of 339 critically ill patients (AUC 0.83 for 14-day mortality), but the number of patients reaching the mortality endpoint was small (n¼14).11 The ARDS substudy reported an independent association with IL-18 to 28-day mortality (IL-18 .200 pg ml21, hazards ratio 2.32, 95% CI 1.2 –4.4).13 Furthermore, a recent study with serum IL-18 suggested that IL-18 is associated with long-term mortality when measured at RRT initiation.28 In summary, no conclusive data of IL-18 for predicting mortality exist. Data from two previous studies suggest that IL-18 levels are higher in patients with sepsis.17 29 However, our sensitivity analysis excluding patients with sepsis did not significantly change the predictive value of IL-18 (Table 3). Furthermore, we constructed multivariable logistic regression models for AKI and 90-day mortality as outcomes. In the model for AKI, IL-18 was an independent factor but judging from the almost nonaltered model-based AUCs with and without IL-18 and the overall poor AUC of the model (0.697 95% CI 0.653 –0.741 with IL-18) this did not contradict our findings. We found that the admission IL-18 was inferior compared with admission NGAL in prediction of all the endpoints (Fig. 3). In AKI and 90-day mortality as the endpoints both IL-18 and NGAL yielded to poor results in this study. In prediction of RRT NGAL was superior compared with IL-18 with an AUC (0.827) that can be categorized as good. However, RRT initiation as an endpoint is complicated: it lacks unified criteria and the benefit of RRT in various clinical situations is still without conclusive proof.30 Corresponding results concerning comparison of IL-18 and NGAL were previously reported regarding AKI.17 We also found that the combination of IL-18 and NGAL was not better in prediction of AKI, which is in agreement with a previous study comprising 451 patients.17 Our study has some important limitations to consider. First, we did not analyse urine samples from all patients of the FINNAKI study, but chose a convenience sample cohort for this substudy blinded to patients’ characteristics and outcome. Given that our sample size is large and adequate and the patients included did not differ in characteristics from those who were not included we consider our population representative without any plausible selection bias. Secondly, attributable to logistic reasons, we did not centrifuge the urine samples before freezing. However, data show that this does not impact
BJA
Nisula et al.
Table 2 Multivariable logistic regression models for AKI and 90-day mortality including all variables tested in univariable models. Odds ratios (ORs) with 95% CI are given for variables proved significant in multivariable models (P,0.05). 95% CI, 95% confidence interval; AKI, acute kidney injury; corticosteroids (permanent medication); CKD, chronic kidney disease; colloids, starch or gelatin; COPD, chronic obstructive pulmonary disease; IL-18 max, highest IL-18 during first 24 h in the ICU; ns, not significant (P.0.05); NSAID, non-steroid anti-inflammatory drug (permanent medication); OR, odds ratio; RRT, renal replacement therapy; SAPS II, Simplified Acute Physiology Score; SOFA, Sequential Organ Failure Assessment
Age
AKI
90-day mortality
ns
1.045 (1.030 –1.060)
Gender (male)
ns
Diabetes (any) CKD
2.155 (1.029 – 4.510)
ns
COPD
ns
ns
Liver disease
5.305 (2.350 –11.977)
Systolic heart failure
ns ns
ns
Atherosclerosis
ns
ns
NSAID
ns
Corticosteroids
ns
ns
Pre-ICU diuretics
ns
ns
Pre-ICU colloids
ns
Pre-ICU hypotension
ns
ns
Non-operative admission
ns
Emergency surgery (,1 week)
ns
SAPS II (-age minus renal points)
0.974 (0.956 – 0.992)
SOFA (Day 1 min renal points)
1.160 (1.067 – 1.260)
1.060 (1.040 –1.080) ns
Highest lactate (Day 1)
1.103 (1.026 – 1.186)
ns
Severe sepsis
1.687 (1.155 – 2.466)
ns
Acute liver failure
ns
2.748 (1.026 –7.359)
AKI
–
2.064 (1.390 –3.066)
RRT
–
ns
IL-18
1.003 (1.001 – 1.005)
ns
Model-based ROC AUC without IL-18
0.693 (0.649 – 0.737)
0.824 (0.795 –0.854)
Model-based ROC AUC with IL-18
0.697 (0.653 – 0.741)
0.824 (0.795 –0.854)
Table 3 Highest urine IL-18 of 24 h in prediction of AKI, RRT, and 90-day mortality separately. 95% CI, confidence interval; AUC, area under receiver characteristics curve; AKI, acute kidney injury; RRT, renal replacement therapy; LR+, positive likelihood ratio; N/A, not available (numbers inadequate for calculations) Sensitivity
Specificity
AUC (95% CI)
Cut-off value (pg ml21)
LR1
All study patients (n¼1439) New AKI (n¼268)
0.384
0.778
0.586 (0.546 –0.627)
65
1.72 (1.41 – 2.08)
RRT (n¼49)
0.551
0.739
0.655 (0.572 –0.739)
65
2.04 (1.54 – 2.69)
90-day mortality (n¼289)
0.398
0.683
0.536 (0.497 –0.574)
44
1.25 (1.06 – 1.47)
Non-septic patients (n¼749) New AKI (n¼116)
0.319
0.838
0.563 (0.502 –0.624)
65
1.92 (1.38 – 2.67)
RRT (n¼16)
0.688
0.806
0.780 (0.668 –0.893)
65
N/A
90-day mortality (n¼98)
0.306
0.802
0.534 (0.472 –0.597)
64
1.53 (1.10 – 2.14)
Patients with known baseline creatinine (n¼917) New AKI (n¼174)
0.328
0.805
0.570 (0.519 –0.620)
85
1.68 (1.129 –2.20)
RRT (n¼32)
0.563
0.698
0.643 (0.539 –0.748)
54
1.86 (1.35 – 2.57)
90-mortality (n¼212)
0.703
0.524
0.552 (0.508 –0.597)
4
1.11 (1.00 – 1.23)
on the stability of IL-18.31 Thirdly, in some studies the levels of IL-18 were normalized to urine Cr levels, and the recent meta-analysis suggested that normalized values improved the
Page 6 of 9
performance of IL-18 slightly.18 However, this finding is controversial, because the urine creatinine excretion rates are highly variable.32 Thus, we chose to report the results as absolute
Downloaded from http://bja.oxfordjournals.org/ at Laurentian University on December 5, 2014
Hypertension
BJA
Interleukin-18 and acute kidney injury
A
B IL-18 NGAL IL-18*NGAL
1.0
Sensitivity
0.6
0.4
IL-18 NGAL IL-18*NGAL
0.6
0.4
0.0
0.2
0.4
0.6
0.8
1.0
0.0
0.2
0.4
0.6
0.8
1.0
0.0
0.2
RRT
0.6
0.8
1.0
90-day mortality
IL-18
0.531 (0.479–0.584)
IL-18
AUC (95% CI) 0.598 (0.498–0.697)
IL-18
AUC (95% CI) 0.524 (0.476–0.573)
NGAL
0.631 (0.579–0.682)
NGAL
0.827 (0.765–0.889)
NGAL
0.618 (0.573–0.664)
IL-18* NGAL
0.603 (0.553–0.654)
IL-18* NGAL
0.767 (0.708–0.826)
IL-18* NGAL
0.588 (0.542–0.634)
Fig 3 Receiver operating characteristics curves (ROC AUC) (with 95% CI) of admission urine IL-18, admission urine NGAL, and the combination of urine IL-18 and NGAL in the prediction of AKI, RRT, and 90-day mortality in critically ill patients (n¼855).
concentrations without normalization to creatinine. Fourthly, we had the baseline creatinine available from only 64% patients, and we were obliged to use the recommended19 MDRD equation (with its limitations) to estimate the baseline value if missing. According to the literature, this may lead to over- or underestimation of AKI.33 – 35 Furthermore, to decrease any bias induced by the use of MDRD, we performed a sensitivity analysis excluding patients lacking the baseline creatinine, which procedure did not affect our results. Thus, we consider our findings to be reliable and also generalizable because of the large random sample from consecutive critically ill patients. An obvious strength of our study is that it was a large, prospective, multicentre study evaluating the usefulness of urine IL-18 in the prediction of multiple clinically important patientcentred outcomes (AKI, RRT, 90-day mortality) in critical care. Thus, we consider our findings to be reliable and also generalizable. Moreover, we aimed to construct a plausible design to assess the predictive properties of this biomarker using admission and 24 h samples, a reasonably short 3-day observation period for AKI and RRT with exclusion of patients with AKI or RRT already on admission. Finally, our AKI classification used the latest KDIGO criteria5 including both daily creatinine and hourly urine output measurements. To conclude, shown by the only poor-to-moderate AUCs and fortified by the unaltered predictive model-based AUCs IL-18 has inadequate properties to predict AKI, RRT, or 90-day mortality in critically ill patients.
Authors’ contributions S.N.: drafted the manuscript, performed the statistical analyses, and participated in the design and data gathering of the study. R.Y.: performed the laboratory analyses and revised the
manuscript. M.P.: participated in the data gathering of the study and revised the manuscript. S.T.V.: participated in the design and data gathering of the study and helped to draft the manuscript and to perform the statistical analyses. K.M.K.: participated in the design and coordination of the study, and revised the manuscript. M.T.: participated in the data gathering of the study and revised the manuscript. M.H.: participated in the design and data gathering of the study and revised the manuscript. J.T.: participated in the design of the study and coordination of the laboratory analyses and revised the manuscript. A.M.K.: participated in the design and co-ordination of the study, collected study data, and helped to draft the manuscript. V.P.: principal investigator of the study and participated in the design and coordination of the study, and helped to draft the manuscript and to perform the statistical analyses.
Acknowledgements We are grateful for all the members of the FINNAKI study group in participating hospitals and Tieto Healthcare and Welfare Ltd for database management.
Declaration of interest S.N., R.Y., M.P., S.T.V., K.M.K., M.T., M.H., J.T., A.M.K., and V.P. declare no conflict of interest.
Funding Clinical Research funding (E.V.O.) TYH 2010109/2011210 and T102010070 from Helsinki University Hospital (V.P.), and grants from the Finnish Society of Intensive Care, the Academy of Finland (K.M.K.), the Juselius Foundation (V.P.) and the Pa¨ivikki and Sakari Sohlberg Foundation (V.P.), and the Finnish Society of Anaesthesiologists (S.N.). Laboratory
Page 7 of 9
Downloaded from http://bja.oxfordjournals.org/ at Laurentian University on December 5, 2014
AUC (95% CI)
0.4
1–Specificity
1–Specificity
1–Specificity AKI
0.4
0.0
0.0
0.0
0.6
0.2
0.2
0.2
IL-18 NGAL IL-18*NGAL
0.8
0.8
Sensitivity
0.8
Sensitivity
C 1.0
1.0
BJA analyses were funded by a grant from the Academy of Finland (K.M.K.) and external funding for Critical Care Medicine Research Group (J.T.).
References
Page 8 of 9
17
18
19 20
21
22
23 24
25
26
27
28
29
30 31 32
33
34
35
biomarkers of nephron damage: a multicenter prospective cohort study. J Am Coll Cardiol 2012; 59: 246–55 Siew ED, Ikizler TA, Gebretsadik T, et al. Elevated urinary IL-18 levels at the time of ICU admission predict adverse clinical outcomes. Clin J Am Soc Nephrol 2010; 5: 1497– 505 Liu Y, Guo W, Zhang J, et al. Urinary interleukin 18 for detection of acute kidney injury: a meta-analysis. Am J Kidney Dis 2013; 62: 1058– 67 Kellum JA, Bellomo R, Ronco C. Acute Dialysis Quality Initiative (ADQI): methodology. Int J Artif Organs 2008; 31: 90– 3 Levey AS, Coresh J, Balk E, et al. National Kidney Foundation practice guidelines for chronic kidney disease: evaluation, classification, and stratification. Ann Intern Med 2003; 139: 137–47 Levy MM, Fink MP, Marshall JC, et al. 2001 SCCM/ESICM/ACCP/ATS/ SIS International Sepsis Definitions Conference. Intensive Care Med 2003; 29: 530– 8 Nisula S, Yang R, Kaukonen KM, et al. The urine protein NGAL predicts renal replacement therapy, but not acute kidney injury or 90-day mortality in critically ill adult patients. Anesth Analg 2014; 119: 95– 102 Ray P, Le Manach Y, Riou B, Houle TT. Statistical evaluation of a biomarker. Anesthesiology 2010; 112: 1023– 40 Metzger J, Kirsch T, Schiffer E, et al. Urinary excretion of twenty peptides forms an early and accurate diagnostic pattern of acute kidney injury. Kidney Int 2010; 78: 1252–62 Arthur JM, Hill EG, Alge JL, et al. Evaluation of 32 urine biomarkers to predict the progression of acute kidney injury after cardiac surgery. Kidney Int 2014; 85: 431–8 Koyner JL, Garg AX, Coca SG, et al. Biomarkers predict progression of acute kidney injury after cardiac surgery. J Am Soc Nephrol 2012; 23: 905– 14 Parikh CR, Mishra J, Thiessen-Philbrook H, et al. Urinary IL-18 is an early predictive biomarker of acute kidney injury after cardiac surgery. Kidney Int 2006; 70: 199–203 Lin CY, Chang CH, Fan PC, et al. Serum interleukin-18 at commencement of renal replacement therapy predicts short-term prognosis in critically ill patients with acute kidney injury. PloS One 2013; 8: e66028 Washburn KK, Zappitelli M, Arikan AA, et al. Urinary interleukin-18 is an acute kidney injury biomarker in critically ill children. Nephrol Dial Transplant 2008; 23: 566–72 Joannidis M, Forni LG. Clinical review: timing of renal replacement therapy. Crit Care 2011; 15: 223 Parikh CR, Butrymowicz I, Yu A, et al. Urine stability studies for novel biomarkers of acute kidney injury. Am J Kidney Dis 2013; 63: 567– 72 Waikar SS, Sabbisetti VS, Bonventre JV. Normalization of urinary biomarkers to creatinine during changes in glomerular filtration rate. Kidney Int 2010; 78: 486–94 Bagshaw SM, Uchino S, Cruz D, et al. A comparison of observed versus estimated baseline creatinine for determination of RIFLE class in patients with acute kidney injury. Nephrol Dial Transplant 2009; 24: 2739– 44 Gaiao S, Cruz DN. Baseline creatinine to define acute kidney injury: is there any consensus? Nephrol Dial Transplant 2010; 25: 3812 – 4 Poggio ED, Nef PC, Wang X, et al. Performance of the CockcroftGault and modification of diet in renal disease equations in estimating GFR in ill hospitalized patients. Am J Kidney Dis 2005; 46: 242– 52
Downloaded from http://bja.oxfordjournals.org/ at Laurentian University on December 5, 2014
1 Bagshaw SM, George C, Dinu I, Bellomo R. A multi-centre evaluation of the RIFLE criteria for early acute kidney injury in critically ill patients. Nephrol Dial Transplant 2008; 23: 1203–10 2 Joannidis M, Metnitz B, Bauer P, et al. Acute kidney injury in critically ill patients classified by AKIN versus RIFLE using the SAPS 3 database. Intensive Care Med 2009; 35: 1692– 702 3 Nisula S, Kaukonen KM, Vaara ST, et al. Incidence, risk factors and 90-day mortality of patients with acute kidney injury in Finnish intensive care units: the FINNAKI study. Intensive Care Med 2013; 39: 420–8 4 Vaara ST, Pettila¨ V, Kaukonen KM, et al. The attributable mortality of acute kidney injury: a sequentially matched analysis. Crit Care Med 2013; 42: 878– 85 5 Kidney Disease: Improving Global Outcomes (KDIGO) Acute Kidney Injury Work Group. KDIGO Clinical Practice Guideline for Acute Kidney Injury. Kidney Inter, Suppl 2012; 2: 1– 138 6 McCullough PA, Shaw AD, Haase M, et al. Diagnosis of acute kidney injury using functional and injury biomarkers: workgroup statements from the tenth Acute Dialysis Quality Initiative Consensus Conference. Contrib Nephrol 2013; 182: 13 –29 7 Dinarello CA, Novick D, Kim S, Kaplanski G. Interleukin-18 and IL-18 binding protein. Front Immunol 2013; 4: 289 8 Melnikov VY, Faubel S, Siegmund B, Lucia MS, Ljubanovic D, Edelstein CL. Neutrophil-independent mechanisms of caspase-1and IL-18-mediated ischemic acute tubular necrosis in mice. J Clin Invest 2002; 110: 1083–91 9 Melnikov VY, Ecder T, Fantuzzi G, et al. Impaired IL-18 processing protects caspase-1-deficient mice from ischemic acute renal failure. J Clin Invest 2001; 107: 1145– 52 10 Parikh CR, Jani A, Melnikov VY, Faubel S, Edelstein CL. Urinary interleukin-18 is a marker of human acute tubular necrosis. Am J Kidney Dis 2004; 43: 405–14 11 Doi K, Negishi K, Ishizu T, et al. Evaluation of new acute kidney injury biomarkers in a mixed intensive care unit. Crit Care Med 2011; 39: 2464– 9 12 Endre ZH, Pickering JW, Walker RJ, et al. Improved performance of urinary biomarkers of acute kidney injury in the critically ill by stratification for injury duration and baseline renal function. Kidney Int 2011; 79: 1119– 30 13 Parikh CR, Abraham E, Ancukiewicz M, Edelstein CL. Urine IL-18 is an early diagnostic marker for acute kidney injury and predicts mortality in the intensive care unit. Clin J Am Soc Nephrol 2005; 16: 3046– 52 14 Krawczeski CD, Goldstein SL, Woo JG, et al. Temporal relationship and predictive value of urinary acute kidney injury biomarkers after pediatric cardiopulmonary bypass. J Am Coll Cardiol 2011; 58: 2301–9 15 Zheng J, Xiao Y, Yao Y, et al. Comparison of urinary biomarkers for early detection of acute kidney injury after cardiopulmonary bypass surgery in infants and young children. Pediatr Cardiol 2013; 34: 880–6 16 Nickolas TL, Schmidt-Ott KM, Canetta P, et al. Diagnostic and prognostic stratification in the emergency department using urinary
Nisula et al.
Interleukin-18 and acute kidney injury
Appendix
Central Hospital: Meri Poukkanen, Esa Lintula, Sirpa Suominen. La¨nsi Pohja Central Hospital: Jorma Heikkinen, Timo Lavander, Kirsi Heinonen, Anne-Mari Juopperi. Middle Ostrobothnia Central Hospital: Tadeusz Kaminski, Fiia Ga¨ddna¨s, Tuija Kuusela, Jane Roiko. North Karelia Central Hospital: Sari Karlsson, Matti Reinikainen, Tero Surakka, Helena Jyrko¨nen, Tanja Eiserbeck, Jaana Kallinen. Satakunta Hospital district: Vesa Lund, Pa¨ivi Tuominen, Pauliina Perkola, Riikka Tuominen, Marika Hietaranta, Satu Johansson. South Karelia Central Hospital: Seppo Hovilehto, Anne Kirsi, Pekka Tiainen, Tuija Mylla¨rinen, Pirjo Leino, Anne Toropainen. Tampere University Hospital: Anne Kuitunen, Ilona Leppa¨nen, Markus Levoranta, Sanna Hoppu, Jukka Sauranen, Jyrki Tenhunen, Atte Kukkurainen, Samuli Kortelainen, Simo Varila. Turku University Hospital: Outi Inkinen, Niina Koivuviita, Jutta Kotama¨ki, Anu Laine. Oulu University Hospital: Tero Ala-Kokko, Jouko Laurila, Sinikka Sa¨lkio¨. Vaasa Central Hospital: Simo-Pekka Koivisto, Raku Hautama¨ki, Maria Skinnar. Handling editor: J. P. Thompson
Page 9 of 9
Downloaded from http://bja.oxfordjournals.org/ at Laurentian University on December 5, 2014
The FINNAKI study Group Central Finland Central Hospital: Raili Laru-Sompa, Anni Pulkkinen, Minna Saarelainen, Mikko Reilama, Sinikka Tolmunen, Ulla Rantalainen, Marja Miettinen. East Savo Central Hospital: Markku Suvela, Katrine Pesola, Pekka Saastamoinen, Sirpa Kauppinen. Helsinki University Central Hospital: Ville Pettila¨, Kirsi-Maija Kaukonen, Anna-Maija Korhonen, Sara Nisula, Suvi Vaara, Raili Suojaranta-Ylinen, Leena Mildh, Mikko Haapio, Laura Nurminen, Sari Sutinen, Leena Pettila¨, Helina¨ Laitinen, Heidi Syrja¨, Kirsi Henttonen, Elina Lappi, Hillevi Boman. Jorvi Central Hospital: Tero Varpula, Pa¨ivi Porkka, Mirka Sivula Mira Rahkonen, Anne Tsurkka, Taina Nieminen, Niina Prittinen. Kanta-Ha¨me Central hospital: Ari Alaspa¨a¨, Ville Salanto, Hanna Juntunen, Teija Sanisalo. Kuopio University Hospital: Ilkka Parviainen, Ari Uusaro, Esko Ruokonen, Stepani Bendel, Niina Rissanen, Maarit La˚ng, Sari Rahikainen, Saija Rissanen, Merja Ahonen, Elina Halonen, Eija Vaskelainen. Lapland
BJA
