The treatment and prevention of acute kidney injury (AKI) remains a major challenge to the intensive care physician. In June 2017, ICM dedicated a whole issue to this important clinical syndrome which remains associated with increased short- and long-term morbidity and mortality [1]. In the current editorial, we focus on recent papers handling the pathophysiology, prevention and supportive treatment of AKI during critical illness, highlighting publications from ICM in the last year (Table 1).

Table 1 Questions addressed by original trials related to AKI and published in ICM in 2017
Full size table

In recent literature, the importance of contrast-induced AKI is increasingly questioned. This is a particular concern in the critically ill when contrast studies are often required to guide the management of complex patients with or at risk of AKI. A single-center retrospective observational trial including 6877 ICU patients compared the incidence of AKI in patients receiving contrast-enhanced or -unenhanced CT scans in subgroups based on pre-imaging renal function (eGFR < or > 45 ml/min/1.73 m2) [2]. Post-contrast AKI was defined based on both oliguria and creatinine KDIGO criteria in the 72 h after CT scanning. To address selection bias, patients receiving contrast were propensity-matched to those who had unenhanced scans. In 1223 matched pairs with eGFR > 45 rates of AKI, dialysis and mortality were not higher in the contrast group. In 285 matched pairs with eGFR < 45, a mix of patients with pre-existing CKD or existing AKI prior to imaging, rates of AKI did not differ. However, the contrast group had higher rates of dialysis, although overall the incidence of dialysis was low (26 patients overall). While this might reflect true contrast toxicity, the lack of difference in AKI by creatinine or urine output suggests the decision to start dialysis post-CT might also be driven by perceived toxicity of contrast [2]. Similarly, Ehrmann et al. [3] performed a meta-analysis of recent controlled studies comparing matched patients exposed or unexposed to contrast (3 studies with 560 patients) and concluded that contrast administration is not associated with AKI (OR 0.95; 0.45–1.62). Using a Bayesian analysis, the authors demonstrate that to conclude that AKI is attributable to contrast requires a disproportionate and biased weighting of evidence toward older uncontrolled observational studies [3]. These two studies suggest that the risk of contrast-induced AKI in ICU is much lower than previously believed. This is in line with the conclusions from the Expert opinion of the Working Group on Prevention, AKI section, European Society of Intensive Care Medicine suggesting “not delaying urgent contrast-enhanced investigations or interventions for potential preventative measures” [4]. Reflecting this change of emphasis, the terminology contrast-induced AKI has been changed to contrast-associated AKI (CA-AKI) or post-contrast AKI (PC-AKI).

Timely prevention and/or treatment of AKI requires accurate risk prediction or early diagnosis. Using random forest machine-learning techniques applied to a multi-center database of patients from the EPaNIC study, Flechet et al. [5] developed and validated an electronic AKI prediction calculator, made available on a website, with separate models for predicting AKI incidence and severity (using creatinine criteria only). Four models, reflecting progressively increasing data availability over time from baseline (preadmission) to day 1, showed progressively increasing discrimination (AUROC rising from 0.75 to 0.82 for any AKI and from 0.77 to 0.84 for AKI 2–3). Models were well calibrated, and decision curve analysis showed clinical usefulness over a wide range of risk thresholds. Importantly, the model based on data available on ICU admission (age, baseline serum creatinine, diabetes, type of admission, blood glucose, suspected sepsis and hemodynamic support) performed slightly better than serum neutrophil gelatinase-associated lipocalin (NGAL) on admission [5]. Indeed, the clinical value of AKI biomarkers remains controversial [6]. Clinical studies examining the benefits of biomarker-directed management are thus required to establish their clinical role. To this end, in a single center study, Meersch et al. [7] identified 276 elective cardiac surgery patients as at risk of AKI based on the level of a urinary marker of cell cycle arrest (TIM2*IGFBP7) at 4 h after cardiopulmonary bypass. These biomarker-positive patients were randomized to standard care or application of a care bundle for AKI prevention based on the 2012 KDIGO clinical practice guideline for AKI. The intervention significantly reduced the incidence of AKI within the first 72 h post-surgery from 71.7 to 55.1% (p  = 0.004). However, more than 80% of the AKI cases were defined only on urine output criteria, and therefore the effect of the intervention mainly relied on a reduction in oliguria. Furthermore, secondary outcomes (LOS, mortality or major adverse kidney event at 30, 60 and 90 days) were not affected [7]. With these caveats in mind, this is the first study showing the impact of a preventive intervention on the incidence of AKI. The biomarker-selected design excluding patients at low risk may be an important determinant of this positive result [8].

Optimal fluid management in patients with or at risk of AKI represents an everyday clinical problem. In the AKI-themed special issue of ICM, an extensive review on fluid management in AKI discussed the potential pitfalls of using oliguria as a trigger for fluid administration [9], whereas another (short) paper provided valuable guidance regarding the choice between fluids and vasopressor therapy in oliguric patients [10]. In addition to the absolute volume of intravenous fluid, the impact of fluid composition on kidney function has been the subject of several clinical trials over the past decade. The LICRA (Limiting IV Chloride to reduce AKI) study used an open-label four-period design to compare a chloride-limited to a chloride-rich fluid strategy in 1136 patients undergoing cardiac surgery [11]. The primary outcome, the peak delta post-operative creatinine, was not significantly affected. The findings of this open non-randomized study [11] contrast with a recent cluster-randomized, multiple cross-over trial in a large (n = 15,802) general ICU population (SMART trial) [12] which demonstrated a reduction of major adverse kidney events within 30 days with balanced crystalloids compared with saline. When comparing these results to LICRA, it is important to note that, in patients admitted to a cardiac ICU in SMART, no signal for the benefit of buffered solutions was seen.

Finally, best practice supportive care for patients requiring renal replacement therapy (RRT) in the ICU remains an area of ongoing research and debate. Importantly, lacking high-quality evidence derived from randomized trials, numerous features of RRT prescription and delivery are not standardized [13]. Given these concerns, and the difficultly in establishing definitive evidence from clinical trials, there is an increasing interest in quality improvement interventions for the delivery of RRT in the ICU. However, quality improvement requires accepted metrics to judge performance. Quality measures for CRRT were evaluated in a recent systematic review evaluating many aspects of prescription and delivery of CRRT, such as solute removal which were inconsistently evaluated and, consequently, there were no clear benchmarks for CRRT delivery assessment [14].

Several summary statements have been published over the last year reflecting the current state of research and clinical practice. The AKI workgroup of ESICM has updated its guidelines for the prevention of AKI. Although the strength of many recommendations has increased, the absence of high-level evidence for many of the discussed interventions precludes strong positive recommendations, with the most definitive recommendations being against the use of starches, dopamine, and high-dose perioperative statins [4]. Future directions in AKI research were also summarized by a group of experts [15]. They suggest focusing on the pathophysiology of different AKI phenotypes, the issue of early diagnosis and biomarker-driven interventions, organ cross-talk and the optimal (timing of) RRT to improve short- and long-term outcomes and interventions to promote renal recovery [15]. Many studies in these areas are ongoing, with others in development, and we await these developments expectantly, as the burden of illness associated with AKI in the ICU remains high with a lack of proven positive benefit from specific interventions.