Improved creatinine-based early detection of acute kidney injury after cardiac surgery

Abstract

 

OBJECTIVES

This study aims to improve early detection of cardiac surgery-associated acute kidney injury (CSA-AKI) compared to classical clinical scores.

METHODS

Data from 7633 patients who underwent cardiac surgery between 2008 and 2018 in our institution were analysed. CSA-AKI was defined according to the Kidney Disease Improving Global Outcomes (KDIGO) criteria. Cleveland Clinical Score served as the reference with an area under the curve (AUC) 0.65 in our cohort. Based on that, stepwise logistic regression modelling was performed on the training data set including creatinine (Cr), estimated glomerular filtration rate (eGFR) levels and deltas (ΔCr, ΔeGFR) at different time points and clinical parameters as preoperative haemoglobin, intraoperative packed red blood cells (units) and cardiopulmonary bypass time (min) to predict CSA-AKI in the early postoperative course. The AUC was determined on the validation data set for each model respectively.

RESULTS

Incidence of CSA-AKI in the early postoperative course was 22.4% (n = 1712). The 30-day mortality was 12.5% in the CSA-AKI group (n = 214) and in the no-CSA-AKI group 0.9% (n = 53) (P < 0.001). Logistic regression models based on Cr and its delta gained an AUC of 0.69; ‘Model eGFR_CKD-EPI’ an AUC of 0.73. Finally, ‘Model DynaLab’ including dynamic laboratory parameters and clinical parameters as haemoglobin, packed red blood cells and cardiopulmonary bypass time improved AUC to 0.84.

CONCLUSIONS

Model DynaLab’ improves early detection of CSA-AKI within 12 h after surgery. This simple Cr-based framework poses a fundament for further endeavours towards reduction of CSA-AKI incidence and severity.

INTRODUCTION

Cardiac surgery-associated acute kidney injury (CSA-AKI) occurs in up to 30% of the patients and remains a challenge, posing significant morbidity and mortality as well as enormous costs for the healthcare system [1–3]. Patients with preoperatively impaired renal function are at highest risk, but also patients with normal renal function may experience CSA-AKI [4]. Reasons are manifold ranging from patient-specific characteristics to procedure complexity along with various perioperative factors [1, 4, 5]. In the past, efforts have been made to develop scoring systems to predict the risk of CSA-AKI associated renal replacement therapy (RRT). The Cleveland Clinical Score (CCS) [6], the Simplified Renal Index for RRT [7] and the Mehta Score also predicting CSA-AKI [8] are among the most frequently used and validated.

Underlining the importance of clinical factors, these scores inherently lack incorporation of parameters mirroring dynamic changes of kidney function. Thus, the aim of the present study was to develop a new tool for enhanced early postoperative CSA-AKI detection.

METHODS

Study population

In total, 8173 consecutive patients who underwent cardiac surgery with cardiopulmonary bypass (CPB) at the Department of Cardiac Surgery at the Klinikum Nürnberg—Paracelsus Medical University (Nuremberg, Germany) were evaluated between April 2008 and January 2018. Patients with indication for heart transplantation, left ventricular assist device implantation or transcatheter aortic valve procedure were not included in the cohort. Pseudo anonymized data collection was done retrospectively via the clinical information system. From the initial 8173 patients, 540 were excluded to avoid potential biasing preconditions for CSA-AKI:

• RRT preoperatively (n = 340)

• impaired renal function defined as estimated Glomerular Filtration Rate (eGFR) < 30 ml in accordance with the Chronic Kidney Disease Epidemiology Collaboration Estimated Glomerular Filtration Rate [9] (eGFR_CKD-EPI) (n = 144)

• without the preoperative creatinine (Cr) value (n = 23)

• without any postoperative value of Cr because they died during or shortly after the operation (n = 15)

• after kidney transplantation (n = 10)

• without information on preoperative RRT (n = 6)

• dialysis intraoperatively (n = 2)

Totally 7633 patients were finally applicable for analyses.

Ethical statement

The study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki. The Institutional Review Board of the Paracelsus Medical University Nuremberg approved the study (IRB-2020–013) on 23 April 2020. Due to the retrospective character of the study using pre-existing, de-identified data, informed consent was waived.

Creatinine-based parameters

Plasma Cr was measured with a kinetic Jaffe method using the clinical chemistry systems from the AU-series (Beckman Coulter, Krefeld; Germany) or from the cobas c-Series (Roche Diagnostics, Mannheim, Germany).

Preoperative Cr and estimated glomerular filtration rate (eGFR) were taken within 3 days before surgery. Blood sampling was performed immediately at intensive care unit (ICU) arrival. Due to individual decisions of the attending clinicians and after careful revision of the data, we found at least one value for all patients within the time period of 6–12 h. In addition to absolute preoperative and early postoperative Cr and eGFR values, their respective changes (ΔCr and ΔeGFR) were calculated as:

ΔXpostop = (Xpostop—Xpreop)/Xpreop.

In case of multiple samples in given timeframe, the higher value was taken. All values were evaluated with further clinical variables for the detection of outcomes. Regarding eGFR-calculation in Black-African or Afro-American people, we found only 34 patients with this etymological characteristics representing 0.45% of the entire cohort, without any influence on the statistical outcome.

Cardiac surgery-associated acute kidney injury definition

CSA-AKI was defined according to the Kidney Disease Improving Global Outcomes (KDIGO) criteria [10] as an increase in Cr by ≥0.3 mg/dl or by ≥26.5 μmol/l from baseline within 48 h after cardiac surgery or increase in Cr to ≥1.5 times baseline within 7 days after cardiac surgery labelled AKI stage 1. AKI stage 2 was defined with an increase of Cr by 2.0–2.9 times baseline. AKI stage 3 was defined with an increase of Cr by ≥3.0 times baseline or with Cr of ≥353.6 μmol/l or with the initiation of RRT, excluding urine output criteria, respectively. CSA-AKI was further divided into transient, for those who recovered within 48 h and persistent, similar to the definition by Basu et al. [11].

According to recent findings, no consensus exists on the definition of CSA-AKI in the current literature with over 35 different definitions of CSA-AKI [8, 12]. However, most researchers have used the Acute Kidney Injury Network (AKIN) and/or the Risk, Injury, Failure, Loss, End Stage Kidney Disease (RIFLE) criteria to define AKI in the first 48 h after surgery. The application of AKIN and RIFLE criteria might lead to underdiagnosing of CSA-AKI. This is why ‘The Kidney Disease: Improving Global Outcomes’ (KDIGO) criteria exhibiting greater sensitivity to detect AKI with better prediction of associated in hospital mortality have been used in newer studies to better depict the underlying multifactorial processes in background [10]. These criteria, which essentially merge the RIFLE and AKIN criteria, have become the new consensus definition of AKI and can be used to diagnose CSA-AKI invariably including the whole period of 7 days after the surgery.

CCS established by Thakar et al. [6] to estimate the risk of dialysis after cardiac surgery and revalidated also for the risk of severe CSA-AKI was applied as basic clinical score [12–14].

Statistical analysis

The descriptive statistics of preoperative, intraoperative and postoperative data were performed to compare patients with and without CSA-AKI. Categorical variables were summarized as frequencies and percentages, while scalar variables were summarized as median with ranges from first to third quartiles, since the data were not normally distributed. The normality of data was tested with the Shapiro–Wilk test. Fisher’s exact test (in the case of categorical variables with frequencies below 5) or Chi square tests were used to compare patient characteristics, risk status and outcomes between CSA-AKI and non-CSA-AKI groups in the case of categorical variables, while Mann–Whitney U-tests were applied for scalar variables. Repeated-measures analysis of variance was performed to assess the differences in Cr and eGFR. The statistical results and modelling were reported on binary response of the AKI variable, i.e. AKI versus non-AKI. Additionally we divided it into 3 subtypes (stage 1–3) to draw attention to even mild kidney injury.

To evaluate Cr and ΔCr and eGFR and ΔeGFR for early detection of CSA-AKI, the multivariable logistic regression modelling was conducted. All logistic regression models were built by following a stepwise-forward approach with the Akaike Information Criteria for optimal model selection [15]. The models were estimated on the training set and validated on the validation set. Data sets were constructed by randomly divided patients in the ratio 0.8:0.2 between training data set, which included 6109 patients, and validation data sets, which included 1524 patients.

The logistic regression results were reported as odds with 95% confidence intervals, obtained on the training data set, while the receiver operating characteristic analyses were made on the validation data set, where the area under the receiver operating characteristic curve [area under the curve (AUC)], specificity and sensitivity at Youden cut-off points were computed for each model respectively.

All statistical analyses were performed by using R (https://www.R-project.org/, version 3.5.3, The R Foundation for Statistical Computing, Vienna, Austria) and a P-value of <0.05 was considered statistically significant.

RESULTS

In the final study sample, 7633 patients were analysed. Demographics and preoperative characteristics are depicted in Table 1. Perioperative characteristics are shown in Table 2. The time course of Cr-Levels and eGFR is shown in Figs 1 and 2.

Table 1:

Demographics and preoperative clinical status

Characteristic	Missing data	No AKI (n = 5921)	AKI (n = 1712)	P-value
Demographics
Age (year)	0	69.0 [61.0–75.0]	73.0 [65.0–77.0]	<0.001
Female sex	0	1559 (26.3%)	500 (29.2%)	0,02
BMI	0	27.4 [24.9–30.4]	27.4 [24.8–31.0]	0,55
Preoperative information
Medical history
Hypertension	11	5446 (92.0%)	1577 (92.5%)	0.575
Hyperlipidaemia	20	4924 (83.2%)	1332 (78.5%)	<0.001
NIDDM	11	1773 (30.0%)	601 (35.2%)	<0.001
IDDM	11	525 (8.87%)	221 (13.0%)	<0.001
Current or ex-smoker	28	3107 (52.6%)	833 (49.0%)	0.009
COPD	4	764 (12.9%)	304 (17.8%)	<0.001
Compensated kidney disease	0	610 (10.3%)	384 (22.4%)	<0.001
Preoperative infection	6	254 (4.29%)	199 (11.6%)	<0.001
Infectious endocarditis	6	162 (2.74%)	148 (8.66%)	<0.001
Catheter to surgery ≤2 days	284	442 (7.67%)	177 (11.2%)	<0.001
Preoperative laboratory values
Haematocrit [%]	16	40.0 [38.0–43.0]	39.0 [36.0–42.0]	<0.001
Haemoglobin [g/dl]	7	13.8 [12.7–14.8]	13.1 [11.6–14.3]	<0.001
Cr (last value before operation) [mg/dl]	0	0.95 [0.83–1.11]	1.01 [0.85–1.21]	<0.001
eGFR [ml/min/1.73 m²]	0	77.2 [62.8–88.5]	69.0 [54.6–84.6]	<0.001
≥60 ml/min/1.73 m²	0	4707 (79.5%)	1118 (65.3%)	<0.001
<60 ml/min/1.73 m²	0	1214 (20.5%)	594 (34.7%)	<0.001
Preoperative cardiac status
NYHA class III+IV	1	3277 (55.3%)	1117 (65.3%)	<0.001
Atrial fibrillation	2	625 (10.6%)	381 (22.3%)	<0.001
Ejection fraction [%]	13	60.0 [50.0–65.0]	60.0 [47.0–65.0]	<0.001
Myocardial infarction within last 3 weeks	1	970 (16.4%)	340 (19.9%)	0.001
Cardiogenic shock within last 48 h—acute cardiogenic shock	1	97 (1.64%)	103 (6.02%)	<0.001
IABP preoperatively	0	96 (1.62%)	77 (4.50%)	<0.001
Previous cardiac surgery ≥1	0	317 (5.35%)	142 (8.29%)	<0.001
Preoperative risk scores
Euroscore	2	4.43 [2.27–8.96]	8.83 [4.45–17.7]	<0.001
Cleveland Clinical Score	51	2.00 [1.00–3.00]	3.00 [2.00–4.00]	<0.001
0–2	51	3765 (63.9%)	702 (41.6%)	<0.001
3–5	51	1974 (33.5%)	833 (49.3%)	<0.001
6–8	51	149 (2.53%)	143 (8.47%)	<0.001
>9	51	5 (0.08%)	11 (0.65%)	<0.001

Values are presented as n (%) for categorical data or as median with first and third quartiles for scalar values. P-value for comparison of outcomes between groups. For Clevland Clinic Score a point system was utilized: Female gender: 1; Congestive heart failure: 1; Left ventricular ejection fraction <35%: 1; Preoperative use of IABP: 2; COPD: 1; Insulin-dependent diabetes mellitus: 1; Previous cardiac surgery: 1; Emergency surgery: 2; Surgery type: CABG: 0, Valve only: 1, CABG + Valve: 2, Other cardiac surgeries: 2; Preoperative Cr: 1.2 to <2.1 mg/dl: 2, ≥2.1 mg/dl: 5 REF THAKAR + KIES.

AKI: acute kidney injury; BMI: body mass index; CABG: coronary artery bypass grafting; CPB: cardiopulmonary bypass; COPD: chronic obstructive pulmonary disease ; Cr: creatinine; eGFR: estimated glomerular filtration rate; IABP: intra-aortic balloon pump; IDDM: insulin-dependent diabetes mellitus; NIDDM: non-insulin-dependent diabetes mellitus; NYHA: New York Heart Association; pRBC: packed red blood cells; RRT: renal replacement therapy.

Table 2:

Perioperative course

Characteristic	Missing data	No AKI (n = 5921)	AKI (n = 1712)	P-value
Intraoperative information
Type of surgery: CABG	0	3683 (62.2%)	780 (45.6%)	<0.001
Valve	0	1412 (23.8%)	499 (29.1%)	<0.001
CABG + Valve	0	629 (10.6%)	306 (17.9%)	<0.001
Valved conduit aorta ascendens	0	197 (3.33%)	127 (7.42%)	<0.001
Priority: elective	24	4252 (72.0%)	1121 (65.7%)	<0.001
Urgent	24	1439 (24.4%)	408 (23.9%)	0.719
Emergent	24	212 (3.59%)	177 (10.4%)	<0.001
CPB time [min]	58	80.0 [62.0–101]	96.0 [72.0–132]	<0.001
Cross clamp time [min]	71	47.0 [36.0–64.0]	57.0 [42.0–84.0]	<0.001
Deepest temperature [°C]	69	34.5 [34.0–35.0]	34.2 [33.5–34.8]	<0.001
Circulatory arrest	0	15 (0.25%)	29 (1.69%)	<0.001
pRBC intraoperatively (number of units)	0	0.00 [0.00–0.00]	0.00 [0.00–2.00]	<0.001
Postoperative information
IABP postoperatively	0	131 (2.21%)	131 (7.65%)	<0.001
Revision	14	221 (3.74%)	253 (14.8%)	<0.001
Postoperative laboratory values
Cr ICU arrival [mg/dl]	36	0.84 [0.72–0.98]	0.98 [0.81–1.21]	<0.001
Cr <12 h postop [mg/dl]	21	0.91 [0.79–1.06]	1.17 [0.96–1.43]	<0.001
Cr <24 h postop	28	0.93 [0.79–1.09]	1.32 [1.06–1.62]	<0.001
First ΔCr preop/ICU arrival [%]	36	−12.50 [−20.34–4.24]	−4.29 [−14.29–8.75]	<0.001
Second ΔCr preop/12 h [%]	21	−4.53 [−12.50–4.35]	12.9 [0.00–31.0]	<0.001
eGFR ICU arrival [ml/min/1.73 m²]	36	87.0 [74.6–95.1]	71.7 [57.1–85.4]	<0.001
eGFR <12 h postop [ml/min/1.73 m²]	21	80.7 [67.2–90.7]	58.3 [45.6–72.8]	<0.001
eGFR <24 h [ml/min/1.73 m²]	28	79.2 [64.2–91.3]	50.3 [38.3–64.2]	<0.001
First ΔeGFR preop/ICU arrival [%]	36	10.2 [2.76–22.3]	3.88 [−7.06–16.2]	<0.001
Second ΔeGFR preop/6–12 h [%]	21	3.53 [−3.44–13.0]	−12.14 [−25.98–0.00]	<0.001
Outcomes
Last Cr before discharge from clinic [mg/dl]	0	0.95 [0.82–1.12]	1.18 [0.93–1.57]	<0.001
Last eGFR before discharge [ml/min/1.73 m²]	0	76.8 [62.0–89.8]	57.2 [40.0–77.2]	<0.001
Last eGFR before discharge ≥60 ml/min/1.73 m² N/Y	0	4604 (77.8%)	793 (46.3%)	<0.001
<60 ml/min/1.73 m²	0	1317 (22.2%)	919 (53.7%)	<0.001
<15 ml/min/1.73 m²	0	2 (0.03%)	41 (2.39%)	<0.001
RRT postop	9	37 (0.63%)	355 (20.8%)	<0.001
Any postoperative infection	1	751 (12.7%)	627 (36.6%)	<0.001
30-day mortality	0	53 (0.90%)	214 (12.5%)	<0.001

Values are presented as n (%) for categorical data or as median with first and third quartiles for scalar values. P-value for comparison of outcomes between groups. For Clevland Clinic Score a point system was utilized: Female gender: 1; Congestive heart failure: 1; Left ventricular ejection fraction <35%: 1; Preoperative use of IABP: 2; COPD: 1; Insulin-dependent diabetes mellitus: 1; Previous cardiac surgery: 1; Emergency surgery: 2; Surgery type: CABG: 0, Valve only: 1, CABG + Valve: 2, Other cardiac surgeries: 2; Preoperative Cr: 1.2 to <2.1 mg/dl: 2, ≥2.1 mg/dl: [6, 14].

AKI: acute kidney injury; BMI: body mass index; CABG: coronary artery bypass grafting; COPD: chronic obstructive pulmonary disease; CPB: cardiopulmonary bypass; eGFR: estimated glomerular filtration rate; IABP: intra-aortic balloon pump; ICU: intensive care unit; IDDM: insulin-dependent diabetes mellitus; NIDDM: non-insulin-dependent diabetes mellitus; NYHA: New York Heart Association; RRT: renal replacement therapy; Cr: creatinine; pRBC: packed red blood cells.

Figure 1:

Creatinine over time: mean value (error bars represent 95% CI of mean value); P < 0.001 at any time point. AKI: acute kidney injury; Cr: creatinine; ICU: intensive care unit.

Figure 2:

Estimated glomerular filtration rate over time: mean value (error bars represent 95% CI of mean value); P < 0.001 at any time point. AKI: acute kidney injury; CKD-EPI: Chronic Kidney Disease Epidemiology Collaboration; eGFR: estimated glomerular filtration rate; ICU: intensive care unit.

Overall 30-day mortality was 3.5% (267). EuroSCORE estimated mortality was 4.43% [2.27–8.96] for no-AKI and 8.8% [4.45–17] for CSA-AKI. Overall Incidence of CSA-AKI was 22.4% (n = 1712). The 30-day mortality in no-CSA-AKI patients was 0.9% (n = 53) and in CSA-AKI patients, 12.5% (n = 214) (P < 0.001). Occurrence and severity of CSA-AKI are shown for ICU arrival, 24 and 48 h compared to the third to seventh day as well as newly diagnosed CSA-AKI patients in Fig. 3. Over time, stage 1 patients progressed to higher stages of AKI in 33.4%, persisted in the same stage in 37.8% and transient regarding to Basu et al. [11] in 28.8%.

Figure 3:

Occurrence and severity of cardiac surgery-associated acute kidney injury; dark line represents absolute number of patients with first diagnosis of cardiac surgery-associated acute kidney injury. AKI: acute kidney injury; CSA-AKI: cardiac surgery-associated acute kidney injury; ICU: intensive care unit.

Modelling for cardiac surgery-associated acute kidney injury prediction

Models are shown in Table 3. First using the CCS, ‘Model CCS’ served as reference point with an AUC 0.66 (95% CI 0.62–0.69).

Table 3:

All models for cardiac surgery-associated acute kidney injury-detection

Model	Variables in the model (units or classification)		AUC [95% CI]	Sensitivity	Specificity
‘Model CCS’	CCS	Train Set (n = 6109)	0.658 [0.642, 0.674]	0.803	0.425
		Test Set (n = 1524)	0.657 [0.625, 0.690]	0.580	0.628
‘Model eGFRCKD-EPI’	eGFR preop (ml/min/1.73 m²) + eGFR ICU arrival  +  ΔeGFR preop/ICU arrival (%)	Train Set (n = 6109)	0.735 [0.720, 0.751]	0.617	0.733
		Validation Set (n = 1524)	0.727 [0.694, 0.760]	0.552	0.798
‘Model DynaLab’	eGFR preop (ml/min/1.73 m²) + eGFR ICU arrival  +  ΔeGFR preop/ICU arrival (%) + eGFR 12 h  +  ΔeGFR preop/12 h (%) + haemoglobin + pRBC intraop (units) + CPB time (min)	Train Set (n = 6109)	0.844 [0.832, 0.857]	0.730	0.805
		Test Set (n = 1524)	0.836 [0.807, 0.865]	0.735	0.816

The results are presented in terms of AUC with 95% CI for training and validation sets and with optimal sensitivity and specificity values at Youden cut-off points.

For Cleveland Clinical Score a point system was utilized: Female gender: 1; Congestive heart failure: 1; Left ventricular ejection fraction <35%: 1; Preoperative use of IABP: 2; COPD: 1; Insulin-dependent diabetes mellitus: 1; Previous cardiac surgery: 1; Emergency surgery: 2; Surgery type: CABG: 0, Valve only: 1, CABG + Valve: 2, Other cardiac surgeries: 2; Preoperative Cr: 1.2 to <2.1 mg/dl: 2, ≥2.1 mg/dl: 5 [6, 14].

AKI: acute kidney injury; AUC: area under the curve; CABG: coronary artery bypass grafting; CCS: Cleveland Clinical Score; CKD-EPI: Chronic Kidney Disease Epidemiology Collaboration; CPB: cardiopulmonary bypass; eGFR: estimated glomerular filtration rate; Cr: creatinine; ICU: intensive care unit; pRBC: packed red blood cells.

In the second step, we took pre- and postoperative Cr and eGFR values at ICU arrival similar to Karkouti et al. [1] and Ho et al. [16], and the corresponding deltas. ‘Model eGFR_CKD-EPI’ was found to be more precise in predicting CSA-AKI with an AUC of 0.73 (95% CI 0.69–0.76) than a model based on Cr and its delta alone with an AUC of 0.70 (95% CI 0.67–0.74).

In the final model, ‘DynaLab’, the simplest modifiable factors from our demographic findings as preoperative haemoglobin, intraoperative packed red blood cells (units) and cardiopulmonary bypass time (min) were combined with dynamic laboratory findings based on eGRF within 12 h after surgery. Early CSA-AKI detection improved AUC to 0.84 (95% CI 0.81–0.86). Detailed logistic regression is shown in Table 4.

Table 4:

Logistic regression ‘Model DynaLab’ for detection of cardiac surgery-associated acute kidney injury

Logistic regression for ‘Model DynaLab’	Odds ratio	95% CI [,]	P-value
eGFRCKD-EPI preop [ml/min/1.73 m²]	1.088	[1.058, 1.119]	<0.001
eGFRCKD-EPI ICU arrival [ml/min/1.73 m²]	0.973	[0.938, 1.009]	0.135
eGFRCKD-EPI Δpreop/ICU arrival [ml/min/1.73 m²]	1.029	[1.005, 1.053]	0.018
Hb preop [g/dl]	0.869	[0.827, 0.913]	<0.001
eGFRCKD-EPI 12 h postop [ml/min/1.73 m²]	0.911	[0.876, 0.947]	<0.001
eGFRCKD-EPI Δpreop/12 h postop [ml/min/1.73 m²]	0.969	[0.943, 0.995]	0.022
pRBC intraop [number of units]	1.188	[1.107, 1.278]	<0.001
CPB time [min]	1.001	1.000, 1.003]	0.064

CKD-EPI: Chronic Kidney Disease Epidemiology Collaboration; CPB: cardiopulmonary bypass; eGFR: estimated glomerular filtration rate; Hb: haemoglobin; ICU: intensive care unit; pRBC: packed red blood cells.

DISCUSSION

In our study, we found that more than one-fifth of patients developed any form of CSA-AKI. Of those, 35.0% progressed and 37.1% maintained the assumed stage of CSA-AKI in the first 48 h after surgery.

The discrepancy between expected mortality (EuroSCORE) and observed mortality among patients with CSA-AKI was wide, but still within the CI: 8.83 [4.45–17.7]. Nevertheless, the observed notion just underlines the fact that any, even milder form of CSA-AKI contributes to worse outcomes that are not accounted in the current risk models.

Validation of CCS for RRT and severe AKI compared well to the reported values by Thakar and Englberger [6, 17]. However, identically to Birnie et al. [2], CCS alone performed poorly in discrimination of patients who developed CSA-AKI of any grade. In contrast, preoperative kidney function combined with single Cr value at arrival to ICU predicted CSA-AKI in the modest-to-good range. Finally, incorporating dynamic Cr and eGFR changes within 12 h after surgery allowed for good-to-excellent CSA-AKI detection before the stand-alone rise of Cr-values is obvious.

The older studies anticipated that the absolute increase of Cr, used as standard for the diagnosis of AKI, has a delay of 24–48 h [18]. This delivers late information on the extent of kidney injury, especially under the circumstances of perioperative haemodilution [19, 20]. Recent studies increasingly show that detection of postoperative even subclinical Cr changes could enhance early CSA-AKI diagnosis [16, 21]. Karkouti et al. [1] used the threshold of ΔCr postoperative to preoperative ratio of 1.3 to identify postoperative AKI. Further, Ho et al. [16] demonstrated that a 10% early postoperative Cr increase could significantly predict a higher AKI risk compared to a 10% decrease of Cr predicting a lower CSA-AKI risk. Further, an even minimal increase of Cr was associated with a substantial decrease in survival [20].

Our present study extends the previous findings by demonstrating that multiple low threshold ΔCr within the first 12 h carry more prognostic information in contrast to single sampling. In contrast to Karkouti et al. [1], with reported sensitivity of 20%, our model with sensitivity of 73% and specificity of 82% could also be used as an early clinical detection tool.

Subclinical rise in Cr after cardiac surgery has been originally used to demonstrate higher mortality in the affected groups [22]. The present study offers an earlier insight into the critical period for CSA-AKI development, demonstrating that already less pronounced early Cr decrease and reciprocal eGFR change due to haemodilution carry important prognostic information.

Data from animal studies indicate that the therapeutic window for renal injury is limited to the first 24–48 h after the onset [1, 23, 24]. In our cohort, we found that as high as 72.1% of all patients developing CSA-AKI could benefit from such an early detection and diagnosis followed by preventive interventions like volume management, maintenance of adequate blood pressure, and judicious avoidance of nephrotoxins [25]. Preoperative anaemia as well as intraoperative blood transfusions increase the risk of CSA-AKI [1, 26, 27]. In line with the previous observations, transfusion of more than 2 units of packed red blood cells significantly increased the odds of CSA-AKI in our study. Anticipating that preoperative haemoglobin per se is a significant independent factor for CSA-AKI, the timing of transfusion and optimal entering haemoglobin remains open dilemma for future considerations.

Endeavours to identify novel biomarkers for early kidney injury detection have recognized neutrophil gelatinase-associated lipocalin and combination of tissue inhibitor of metalloproteinases-2 (TIMP-2) with insulin-like growth factor-binding protein 7 [TIMP-2]•[IGFBP7] among the most promising [28–30], yet both biomarkers perform inferiorly with regard to our ‘DynaLAb’ Model. We strongly believe that a combination of Cr changes and injury biomarker is the way to go in further implementation of the diagnostic paradigm for CSA-AKI [29].

Limitations

Data of our study were collected retrospectively from a single centre. External validation is mandatory and needs to be performed in the future. The broad study period could be a confounding because of change in anaesthesiological or clinical protocols over 10 years. Finally, we excluded patients with eGFR <30 ml/min/1.73 m2 and therefore our results may not be applicable to patients with stages 4 and 5 of chronic kidney disease undergoing heart surgery.

CONCLUSIONS

In cardiac surgery patients, determination of immediate sequential Cr and eGFR implemented with assessment of corresponding subclinical changes facilitates early identification of CSA-AKI, better than the widely known Cleveland Clinical Score. The appealing simplicity, specificity and sensitivity of our enriched ‘DynaLab’ model could be implemented easily at most departments worldwide. The model, especially if integrated in the laboratory medical software as an alert system or provided as Apps, might represent a viable, inexpensive clinical tool.

Funding: This research received no external funding.

Conflict of interest: none declared.

Author contributions

Ferdinand Aurel Vogt: Conceptualization; Formal analysis; Methodology; Validation; Visualization; Writing—original draft; Writing—review & editing. Janez Zibert: Data curation; Formal analysis; Software; Validation; Visualization. Alenka Bahovec: Data curation; Formal analysis; Investigation; Project administration; Validation; Visualization. Francesco Pollari: Data curation; Formal analysis; Resources; Visualization. Joachim Sirch: Data curation; Resources; Software; Validation. Matthias Fittkau: Supervision; Writing—review & editing. Thomas Bertsch: Methodology; Resources; Supervision; Writing—review & editing. Martin Czerny: Supervision; Writing—review & editing. Giuseppe Santarpino: Conceptualization; Writing—review & editing. Theodor Fischlein: Supervision; Writing—review & editing. Jurij M. Kalisnik: Conceptualization; Data curation; Formal analysis; Methodology; Supervision; Writing—original draft; Writing—review & editing.

Reviewer information

Interactive CardioVascular and Thoracic Surgery thanks Heyman Luckraz, Milan Milojevic and the other anonymous reviewer(s) for their contribution to the peer review process of this article.

ABBREVIATIONS

AUC

Area under the curve

CCS

Cleveland Clinical Score

CKD-EPI

Chronic Kidney Disease Epidemiology Collaboration

Cr

Creatinine

CSA-AKI

Cardiac surgery-associated acute kidney injury

eGFR

Estimated glomerular filtration rate

Hb

Haemoglobin

ICU

Intensive care unit

KDIGO

Kidney Disease: Improving Global Outcomes

pRBC

Packed red blood cells

RRT

Renal replacement therapy