Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Introduction

Acute kidney injury (AKI) represents an abrupt deterioration in kidney function and is a frequently encountered phenomenon in hospitalized patients [1, 2]. The impact of AKI is most profound among patients admitted to intensive care units (ICU). The incidence of AKI among hospitalized and critically ill patients appears to be increasing. This is likely attributable to demographic transition (i. e., older patients; more comorbid illness; greater prevalence of chronic kidney disease) coupled with admissions associated with increasingly complex medical and/or surgical diseases and the need for multi-faceted diagnostic and therapeutic interventions. The development of AKI now complicates the course in an estimated two-thirds of critically ill patients [3–5]. For those with more severe forms of AKI, an estimated 50–70 % will require support with acute renal replacement therapy (RRT), which represents a small, but important subgroup of all critically ill patients (4–8 %) [6]. The initiation of RRT can contribute to a considerable escalation in both the complexity of illness and associated costs of care. Indeed, these critically ill patients experience substantial morbidity, including non-recovery of kidney function and long-term chronic kidney disease or dialysis dependence [7–9] as well as excess mortality, with hospital mortality rates commonly exceeding 60 % [6, 8].

Classification of AKI

The diagnostic classification of AKI has evolved considerably in the last several years. In 2004, a consensus definition, the RIFLE classification scheme (Acronym: Risk, Injury, Failure, Loss and End-stage renal disease), was published by the Acute Dialysis Quality Initiative (ADQI) group [10]. This classification defined three grades of severity of AKI (Risk, Injury, Failure), based on relative changes to serum creatinine (sCr) and changes in urine output, and two outcome categories (Loss and End-Stage Kidney Disease) based on the duration of RRT. This novel classification scheme has been validated and widely integrated into the medical literature. The RIFLE classification was later refined by the Acute Kidney Injury Network (AKIN), a consortium uniting representatives from all major nephrology and critical care societies. This adaption specified a 48-hour timeline during which the acute changes in kidney function had to occur to fulfill the definition of AKI. The AKIN modification also simplified the stages of severity of AKI, with the first two representing worsening renal dysfunction and the third stage characterized by kidney failure and encompassing any patients initiated on RRT. The RIFLE/AKIN classification schemes have now been consolidated into the KDIGO Clinical Practice Guidelines for Acute Kidney Injury definition for AKI [11] (Table 1).

Table 1 Acute kidney injury (AKI) definition (KDIGO Clinical Practice Guidelines for Acute Kidney Injury) [11]
Full size table

Although the development of this consensus classification scheme has been an important landmark for clinical care and research in AKI, it has recognized limitations. The integration of urine output criteria may be confounded by numerous factors unrelated to the severity of underlying kidney injury (e.g., inability to measure outside of intensive monitoring settings; diuretic therapy). In addition, the stages associated with severity of urine output and sCr criteria may not be aligned in terms of prognosis (e.g., Stage 2 by sCr criteria associated with higher mortality than Stage 2 by urine output criteria). Likewise, the use of sCr has limitations. Because sCr is a surrogate marker of kidney function and not a biomarker of kidney damage, changes to sCr may occur long after the injury stimulus has occurred (i. e., changes in sCr after injury may not rise significantly until there is a 50 % decline in glomerular filtration rate [GFR]) [12]. Serum creatinine is known to be influenced by a number of non-renal factors including patient age, sex, muscle mass, nutritional status and metabolism, medications, and extravascular fluid balance. Finally, baseline sCr values are often unknown (in ~ 30 % of ICU patients). This presents challenges for interpretation and attributing an elevated sCr solely to AKI rather than representing pre-existing chronic kidney disease or a combination of both.

Accordingly, the integration of novel biomarkers that may be sensitive and specific to the detection of early and acute kidney damage would be welcome and are likely essential to appropriately identify, triage and intervene in patients with AKI at increased risk for less favorable outcomes. Similar to conventional markers, such as sCr, these novel biomarkers need to be minimally invasive, easily detectable, reliable and inexpensive, but also need to be a direct biomarker of kidney damage (not a surrogate for function) and show correlation with clinically meaningful and potentially actionable outcomes, such as worsening AKI, need for initiation of RRT, non-recovery of function or a composite of these [13].

Biomarkers in AKI

The concept of novel biomarkers for the accurate and timely diagnosis of AKI brings much promise. However, as with any new information available to the clinician, we must ensure that it will positively inform our decision making process, and not simply provide more ‘noise’ for the clinician to sift through when trying to reach a diagnostic conclusion. For example, when deciding on whether to initiate RRT, the decision is rarely made based solely on biochemical grounds and rather is more likely to be the integration of a number of variables in the setting of a deteriorating clinical picture. It is with all likelihood that biomarkers will eventually play an increasing role in the diagnosis and management of AKI, but integrated into existing decision-making tools and clinical judgment (Table 2). In August 2011, the Acute Dialysis Quality Initiative (ADQI) held a multi-disciplinary consensus conference specifically focused on the evaluation of biomarkers in acute kidney injury (http://www.adqi.net/).

Neutrophil Gelatinase-Associated Lipocalin

Neutrophil gelatinase-associated lipocalin (NGAL) is a 25 kDa protein bound to gelatinase and is typically expressed in several human tissues including the kidney, trachea, lungs, stomach and colon [13, 14]. NGAL is one of the first and most rapidly upregulated transcripts in renal tubular epithelial cells following acute injury [14–16]. The protein by-product of this gene upregulation, NGAL, is easily measured in the urine and plasma, where its concentration increases in a dose-dependent relationship with the severity and duration of acute tubular injury. Numerous observational studies have evaluated urine and serum NGAL for the diagnosis and prognosis of AKI in adult critically ill patients and will be the primary focus of this review (Table 3).

Table 2 A summary of potential actions that clinicians could take in response to elevated kidney-damage biomarkers, such as NGAL
Full size table

Diagnosis of AKI

In pediatric populations, NGAL has been shown to be an early marker of kidney damage, rising significantly as early as two hours after cardiac surgery with cardiopulmonary bypass (CPB) and predictive of development of conventionally defined AKI [17, 18]. However, translation of these findings to adult populations has been variable [12]. This is likely attributable to the added confounding influences in adult patients, such as burden of comorbid illness (e.g., cardiovascular disease, chronic kidney disease) and the multiple heterogeneous contributing factors for AKI [14]. Several studies have examined the validity of NGAL for the diagnosis of AKI in adult critically ill patients [19–23]. Cruz et al. evaluated serum NGAL in 301 consecutive patients admitted to a general medical/surgical ICU. AKI, defined by the RIFLE criteria, occurred in 133 patients (44 %). NGAL was higher in those developing AKI (185 ng/ml vs. 82 ng/ml, p < 0.001) and had good

Table 3 Selected prospective cohort studies examining NGAL in adult patients (studies inclusive of > 100 patients)
Full size table

discrimination for predicting AKI in the subsequent 48 hours (area under receiver operative characteristic curve [AUROC] 0.78), implying a lead-time benefit for the diagnosis of those at elevated risk for AKI [19]. In another large single-center prospective observational study, de Geus et al. [20] measured urine and serum NGAL at ICU admission in 632 consecutive critically ill patients to evaluate its utility to predict development of AKI, defined by the RIFLE criteria, within 1 week. In this study, the discrimination of urine and serum NGAL were higher for more severe AKI (AUROC for RIFLE – Failure 0.88 and 0.86, respectively). Whereas crude prediction was similar to that of admission estimated GFR (eGFR, AUROC 0.84), the integration of urine and serum NGAL in the best clinical prediction model improved discrimination. In addition, urine NGAL was superior to serum NGAL for discriminating between the development of transient and sustained AKI (AUROC 0.80) and added net discrimination improvement to the best clinical prediction model [24]. In another prospective observational study of several novel biomarkers including urine NGAL, in which 529 critically ill patients admitted to two ICUs were enrolled, the diagnostic performance of urine NGAL was improved when stratified by baseline kidney function and time from onset of injury [21].

Subclinical AKI

In a recent pooled analysis of 10 prospective observational studies, detectable elevations in urine and/or plasma NGAL, in the absence of detectable changes in sCr fulfilling the conventional diagnosis of AKI, identified a unique cohort of patients with ‘subclinical AKI’ [25]. These patients were at greater risk of adverse events, a more complicated clinical course and less favorable clinical outcome, including RRT initiation, higher mortality, longer durations of stay in the ICU and hospital when compared to those without evidence of detectable elevations in NGAL. These findings were consistent regardless of whether NGAL was measured in the urine or serum. It has also recently been suggested that NGAL be used as a marker of kidney damage without evident dysfunction and concomitant elevations in sCr in patients undergoing radiographical procedures involving contrast dyes to better define and treat contrast-induced nephropathy [26]. These observations imply the existence of a clinically important state of subclinical kidney ‘damage’ detectable only by injury-specific biomarkers, such as NGAL, that occurs without significant loss to GFR and that may be completely missed by our current diagnostic paradigm for AKI [27].

Prognosis in AKI

NGAL has been investigated in numerous studies across a range of clinical settings (e.g., ICU, emergency department [ED], postoperative) as a prognostic biomarker to discriminate and predict clinically important outcomes. A large systematic review and meta-analysis showed that NGAL had value for predicting not only the development of AKI, but also RRT initiation and mortality [28]. Two studies have recently shown that urine NGAL has high predictive value for the diagnosis of AKI within 72 hours for patients presenting to the ED [22, 29]. Indeed, a single elevated urine NGAL > 130 ng/mg was associated with a 24-fold increased risk of treatment escalation, including nephrology consultation, RRT initiation or ICU admission [29]. In the critically ill, several studies have shown that an elevated NGAL at the time of ICU admission predicts RRT initiation (AUROC 0.82–0.88); prolonged stay in ICU and hospital; and/or increased risk of death; however, prediction of mortality was only fair (AUROC 0.68) [19, 20, 30]. Plasma NGAL has also been shown to correlate with the severity and duration of AKI, as well as overall illness severity (e.g., Acute Physiology and Chronic Health Evaluation II [APACHE II] score), implying it may have value as a valuable global biomarker of prognosis [19]. In addition, when measured in critically ill patients with severe AKI started on RRT, serum NGAL concentrations at the time of RRT initiation were higher in non-survivors compared to survivors, and independently predicted 28-day survival [31]. Elevated NGAL after cardiac surgery not only predicts the development of AKI, but is also associated with improved discrimination for hospital stay, RRT initiation or death (AUROC 0.75). Serum NGAL has also been shown to have value for prediction of worsening or progressing of AKI in the postoperative period [32]. Recently, two studies have shown that NGAL may have value for predicting kidney recovery after an episode of severe AKI [33, 34]. Recovery may depend on a number of independent factors, including baseline kidney function, the severity of the initial insult and whether there are ongoing insults. In a small substudy of critically ill patients from the ATN trial, serial measures of a composite of biomarkers, including NGAL, cystatin C, and hepatocyte growth factor (HGF), when integrated with the best clinical prediction model, showed excellent discrimination for recovery to dialysis independence (AUROC 0.90), implying incremental value for integrating a panel of biomarkers for predicting renal recovery [34].

Conclusion

AKI remains a common and challenging clinical problem for clinicians. Indeed, AKI may be largely iatrogenic and may be partly attributable to the advent of modern ICU support. Importantly, its incidence continues to rise, its associated outcomes are poor, and there are currently few, if any, preventative and/or therapeutic interventions that can modify this prognosis once established. Currently, the standard tools we have to diagnose and monitor AKI (i. e., sCr and urine output) have remained unchanged for several decades and are clearly suboptimal. There are now numerous conventional and novel kidney-injury specific biomarkers that have been characterized and are undergoing clinical validation for the diagnosis and clinical outcome prediction across a spectrum of at-risk patients. This list also includes a reappraisal of conventional measures of kidney function, such as urine output, sCr, and urine microscopy; but also importantly highlights a growing list of novel biomarkers, such as interleukin-18, kidney injury molecule-1, L-type fatty acid binding protein, HGF, N-acetyl-β-D-glucosaminidase, α-glutathione S-transferase, π-glutathione S-transferase and matrix metalloproteinase-9. NGAL remains one of the most well-described biomarkers of kidney damage. While numerous ongoing clinical studies are pursuing further clinical validation, available data imply that this biomarker holds promise to better inform on the early diagnosis and provide important prognostic information (e.g., worsening AKI, need for RRT, renal recovery), in particular when integrated with clinical prediction models.