Secondary Logo

Journal Logo

Predictive models for acute kidney injury after cardiac surgery

Meroni, Roberta; Simonini, Marco; Lanzani, Chiara; Bignami, Elena

European Journal of Anaesthesiology (EJA): January 2018 - Volume 35 - Issue 1 - p 63–65
doi: 10.1097/EJA.0000000000000651
Correspondence
Free

From the Department of Anaesthesia and Intensive Care, IRCCS San Raffaele Scientific Institute (RM, EB), and Genomics of Renal Diseases and Hypertension Department, Vita-Salute San Raffaele University, Milan, Italy (MS, CL)

Correspondence to Elena Bignami, MD, Department of Anaesthesia and Intensive Care, IRCCS San Raffaele Scientific Institute, Via Olgettina 60, 20132 Milan, Italy Tel: +39 02 2643 4524; fax: +39 02 2643 7178; e-mail: bignami.elena@hsr.it

Editor,

We read with interest the work by Echarri et al.,1 and we thank the authors for drawing attention to cardiac surgery-associated acute kidney injury.

Cardiac surgery-associated acute kidney injury is associated with increased risk of mortality and morbidity and predisposes patients to longer hospital stays and increased costs. Despite advances in surgical techniques, in recent years there have been few changes in the prevention, diagnosis and treatment of acute kidney injury. Thus, we think that the main issues concerning this topic remain: early diagnosis and identification of milder forms of acute kidney injury (not requiring renal replacement therapy).1–3

In this prospective, observational, multicentre study, the authors’ aim was to validate previously published predictive models for both dialysis requirement and for a composite endpoint of dialysis or a doubling of serum creatinine level. These models were proposed to analyse both preoperative and intraoperative variables.1 Echarri et al.1 analysed data of 1014 patients. They concluded that these models were able to distinguish patients whose kidney function might worsen after surgery and might probably identify those more prone to postoperative dialysis. However, they underlined the lack of calibration and the difficulty of use, because of the large number of variables considered.1 Moreover, even though lots of early and specific biomarkers (for instance neutrophil gelatinase-associated lipocalin) have been proposed as diagnostic tools during recent years, acute kidney injury diagnostic algorithms (risk, injury, failure, loss, end stage), acute kidney injury network, kidney disease: improving global outcomes (KDIGO) are substantially based on urinary output and serum creatinine, both leading to late diagnosis.1–4

For these reasons the diagnosis of early cardiac surgery-associated acute kidney injury and the identification of high-risk patients still remains a main challenge. Remarkably, recent works still focus on the importance of clinical variables as diagnostic tools and prognostic predictors, as more powerful tools are not yet available.

A recent observational work by Petäjä et al.3 demonstrated the independent association between KDIGO criteria for acute kidney injury, especially urinary output, and 2.5 years mortality. Preoperative assessed mortality risk through Euroscore II was not able to predict acute kidney injury-associated mortality.3 The authors stress the key role of urinary output as an independent mortality predictor and advocate the need of novel prognostic tools focused on clinic variables.

In 2014, in an interesting retrospective work, Birnie et al.5 developed and validated two new risk scores for cardiac surgery-associated acute kidney injury. More than 30 000 patients were included. The KDIGO definition of acute kidney injury was used. Considering preoperative patients characteristics, the authors developed any-stage acute kidney injury risk score and a second risk score only for the development of stage-3 acute kidney injury. Furthermore, they compared them with four previously published scores (Euro, Metha, Cleveland Clinic and Ng scores). We find the results of this study extremely valuable. First, the authors demonstrated the prognostic utility of the KDIGO definition of acute kidney injury, including milder forms of acute kidney injury. Second, they built a preoperative score for any-stage acute kidney injury with better discrimination power [area under the curve (AUC) 0.74] than some of the previously validated scores (such as Euroscore and Cleveland Clinic score) and with an improved calibration. Last, they built a tool that was also able to detect stage-1 acute kidney injury. We emphasise that, even though this work is of extreme significance, the authors were using diagnostic criteria of acute kidney injury that were based on urinary output and serum creatinine, because of missing earlier markers of acute kidney injury.

To improve the prediction of cardiac surgery-associated acute kidney injury, in 2014 we published a new clinical prediction model [CLIN-RISK (a risk model based on clinical variables)], based on data from 802 patients undergoing cardiac surgery.2 We considered clinical variables and used the Northern New England Cardiovascular Disease Study Group score as a reference.2 This was the only risk prediction model focusing on acute kidney injury not requiring dialysis, based exclusively on preoperative data. Primary outcome was the development of stage II or III acute kidney injury (according to the acute kidney injury network). First, we confirmed that the CLIN-RISK model was effective, easy to use and more powerful in a population where the Northern New England Cardiovascular Disease Study Group has previously been validated (AUC 0.79 vs. 0.74, P < 0.01). Moreover, we measured endogenous ouabain in the plasma preoperatively and, to estimate its predictive impact, we added it to the two models. Interestingly, we found that both clinical models were improved by the addition of plasma values of endogenous ouabain (AUC, respectively 0.83 and 0.79).2 This finding is valuable, as preoperative elevated plasma concentrations of endogenous ouabain can identify patients with subclinical kidney injury, hence, who are at higher risk of cardiac surgery-associated acute kidney injury. On this topic, we had previously prospectively studied preoperative plasma concentrations of endogenous ouabain in patients undergoing cardiac surgery and found that it was a powerful and promising biomarker of acute kidney injury and further postoperative complications.4

We stress also that the majority of predictive models focus on identifying patients that would need renal replacement therapy, conversely preoperative identification of patients that suffer from milder forms of acute kidney injury still represents an issue.1,2 Our CLIN-RISK model and the model by Birnie et al.5 both focus on detecting this specific group of patients.2

Another remarkable risk score has been published in 2015 by Jorge-Monjas et al.6 using data of more than 900 patients. Their model [CRATE (CReatinine, lactic Acid, cardiopulmonary bypass Time, EuroSCORE)], that is based on previously demonstrated predictors of acute kidney injury (creatinine and lactic acid at ICU admission, duration of cardiopulmonary bypass and Euroscore), is simple to use and reliable (AUC 0.81). Moreover, it may be performed early after ICU admission, allowing clinicians to start interventions to prevent or reduce the development of acute kidney injury. Notwithstanding its benefits, we suggest that a limitation of this study could be the lack of stratification for different stages of acute kidney injury.

In 2016, Pannu et al.7 published an intriguing risk prediction model of acute kidney injury requiring dialysis based on a small number of preoperative variables, Notably, it differed from other models because it allows for calculating at the bedside the percentage risk of postoperative dialysis.

In conclusion, as explained by Echarri et al.,1 cardiac surgery-associated acute kidney injury represents a key issue in cardiac surgery and prevention, diagnosis and treatment are far from being conclusive. Lots of risk scores and predictive models have been developed, with differing strengths and reliability. Most of them aim to identify last-stage acute kidney injury and allow a late diagnosis of acute kidney injury. New promising biomarkers (for instance, endogenous plasma ouabain) could lead to more and more precise risk models, stratification and diagnostic and prognostic tools that may be able to be dynamically adapted to different populations, paving the way through the development of a real personalised score of acute kidney injury.2,4

Back to Top | Article Outline

Acknowledgements relating to this article

Assistance with the letter: none.

Financial support and sponsorship: none.

Conflicts of interest: none.

Back to Top | Article Outline

References

1. Echarri G, Duque-Sosa P, Callejas R, et al. External validation of predictive models for acute kidney injury following cardiac surgery. Eur J Anaesthesiol 2017; 34:1–8.
2. Simonini M, Lanzani C, Bignami E, et al. A new clinical multivariable model that predicts postoperative acute kidney injury: impact of endogenous ouabain. Nephrol Dial Transplant 2014; 29:1696–1701.
3. Petäjä L, Vaara S, Liuhanen S, et al. Acute kidney injury after cardiac surgery by complete KDIGO criteria predicts increased mortality. J Cardiothorac Vasc Anesth 2016; Aug 26. [Epub ahead of print].
4. Bignami E, Casamassima N, Frati E, et al. Preoperative endogenous ouabain predicts acute kidney injury in cardiac surgery patients. Crit Care Med 2013; 41:744–755.
5. Birnie K, Verheyden V, Pagano D, et al. Predictive models for kidney disease:improving global outcomes (KDIGO) defined acute kidney injury in UK cardiac surgery. Crit Care 2014; 18:606–625.
6. Jorge-Monjas P, Bustamante-Munguira J, Lorenzo M, et al. Predicting cardiac surgery-associated acute kidney injury: the CRATE score. J Crit Care 2016; 31:130–138.
7. Pannu N, Graham M, Klarenbach S, et al. A new model to predict acute kidney injury requiring renal replacement therapy after cardiac surgery. CMAJ 2016; 188:1076–1083.
© 2018 European Society of Anaesthesiology