Abstract
Several biomarkers of renal injury have been identified but the utility of these biomarkers is largely confined to research studies, whereas widespread clinical applicability is limited. This is partly because the use of serum creatinine as the comparator has several limitations and restricts the full interpretation of biomarker performance. To highlight the potential for clinical application of biomarkers, the most pertinent biomarker data are summarized here, using clinically relevant scenarios in which biomarkers could assist with diagnostic and management dilemmas. The paradigms proposed in this review aim to enhance the clinical diagnosis, management, and prognosis of AKI through the combined use of available clinical markers and novel inflammatory, injury, and repair biomarkers.
The field of nephrology has overcome many challenges since Dr. Homer Smith coined the term ARF in 1964.1 In 2002, there were >30 different definitions of ARF in the literature as described by the Acute Dialysis Quality Initiative (ADQI).2 A consensus on a definition was reached in 2004 with the introduction of the term AKI and the use of Risk, Assessment, Failure, Loss, and End Stage Renal Disease. Later, definitions were further refined with the advent of the Acute Kidney Injury Network (AKIN) criteria in 2009 and more recently the Kidney Disease: Improving Global Outcomes (KDIGO) guidelines.3–5 Although we have successfully achieved standardization in AKI screening, the field of nephrology still lacks a more comprehensive definition of AKI that incorporates other dimensions to improve specificity of type, etiology, and prognosis of AKI. Nephrology has yet to overcome its ultimate challenge of solely relying on serum creatinine and occasionally urine output to define AKI and its various causes (Table 1). The limitations of serum creatinine are well known and have been extensively explored in the literature.6–11 The lack of sensitivity of serum creatinine in the detection of AKI can be observed in protocol transplant kidney biopsy specimens, where there is substantial evidence of tubular injury without significant changes in serum creatinine.12 Even when there is a rise in serum creatinine, it occurs 48–72 hours after injury.13 In contrast, several novel biomarkers such as neutrophil gelatinase–associated lipocalin (NGAL) and IL-18 have been shown to increase within hours of injury.14
Definitions of terms used in this review
For this reason, in 2006 the Food and Drug Administration (FDA) released the Critical Path initiative in an effort to develop better evaluation tools such as biomarkers to further characterize diseases such as AKI.15 The FDA defined a “biomarker” as a characteristic that is objectively measured and evaluated as an indicator of normal biologic processes, pathogenic processes, or pharmacologic responses to therapeutic intervention.16 Since commencement of the Critical Path biomarker initiative, there has been tremendous research interest in this area. Over 1000 manuscripts have been published investigating ideal biomarkers to predict early damage before creatinine rise as well as to indicate damage specific to kidney tissue (Figure 1). To date, several novel biomarkers of renal injury have been identified but are largely confined to research studies and have not been widely applied to clinical settings. This is partly due to the use of serum creatinine as the gold standard of AKI diagnosis, which limits the interpretation of biomarker performance to the confines of creatinine use. For example, a biomarker might rise in the setting of true injury but serum creatinine may remain unchanged in that setting due to volume of distribution or compensatory hyperfiltration, causing the biomarker to appear nonspecific. Or a biomarker might not rise as in prerenal azotemia (PRA), which is not true injury to kidney tissue but serum creatinine rises causing the biomarker to appear to lack sensitivity.
Increasing trend in original papers published on renal biomarkers in AKI. The above search was conducted using PubMed (www.ncbi.nlm.nih.gov/pubmed). Mesh terms of AKI and biomarkers were used with the following filters: article type, observational studies, clinical trials; publication dates, 2006–2016 (each year was searched separately); restricted to humans and English language.
Despite the limitations of serum creatinine, certain biomarkers were able to surface as complementary to established clinical markers in assessing patients with AKI. This is in part because novel biomarkers were associated with clinical outcomes such as dialysis and death, which helped address some of the limitations of using serum creatinine as the comparative gold standard.17,18 In an effort to highlight the potential for the clinical application of biomarkers to untangle the web of AKI etiologies and to identify meaningful AKI we will summarize the most pertinent biomarker data using four clinically relevant scenarios in which biomarkers assist with diagnostic dilemmas (Table 2).
Current paradigm versus proposed paradigm in various clinical settings
Scenario 1: Distinguishing PRA from Acute Tubular Injury in Hospitalized Patients
Clinical Dilemma
Both nephrologists and internists frequently face the challenge of differentiating between PRA and acute tubular injury (ATI) when assessing hospitalized patients with AKI. This diagnostic conundrum has been further complicated by the evolving definitions of the two conditions.19 The term PRA is usually used to indicate minimal-to-no structural damage and ATI to indicate true tubular injury. Thus, discerning between these two causes is important because it can alter patient management. If a diagnosis of PRA is made, intravenous fluids (IVF) and optimizing hemodynamics are the staple clinical therapies. In the case of ATI, however, a patient’s condition would not improve with IVFs and unnecessary fluid administration might actually cause harm.20 Often, the diagnosis is confirmed retrospectively rather than prospectively, because clinicians are able to differentiate between the two causes on the basis of the clinical course and duration of creatinine elevation, as well as response to fluids.21,22 Distinguishing between PRA and ATI becomes even more challenging in the setting of preexisting CKD. This scenario is clinically important because AKI superimposed on CKD is associated with worse clinical outcomes.23
Current Paradigm
Clinicians often rely on fractional excretion of sodium (FeNa) to diagnose PRA because it has been shown that FeNa<1% is diagnostic of PRA.24 However, many factors can affect the accuracy of FeNa measurement, because its measurement should be in the absence of diuretics, and precede any IVF administration.25 More so, urinary sodium is affected by multiple nonrenal factors and FeNa<1% has also been described in the setting of ATI.26 Aside from FeNa, urine microscopy findings have also been shown to be strongly diagnostic of ATI versus PRA. However, microscopy use is limited due to several factors such as availability of microscopes in hospital settings, and the fact that few academic clinicians aside from nephrologists are trained in urine microscopy.27 The performance of urine electrolytes and urine microscopy has not been evaluated in AKI etiologies superimposed on CKD.
Proposed Paradigm
The addition of injury biomarkers to the clinical panel could potentially enhance clinician accuracy in distinguishing between PRA and ATI. Injured tubules in ATI elicit inflammatory responses and thus measuring proteins specific to the injured tubule or markers of kidney-specific inflammation in urine would help distinguish ATI from PRA. Urine concentrations of IL-18, which is a mediator of tubular apoptosis, are significantly higher in patients with ATI and longer duration of AKI compared with those with PRA, transient AKI, urinary tract infections, CKD, and healthy control participants.28–31 This protein is robust to sample processing, handling, and storage conditions and provides high discrimination for ATI.32 In addition to IL-18, urine NGAL measured in the emergency department has also been shown to reliably distinguish ATI from PRA.33 Using adjudicated cases of PRA and ATI, NGAL levels >104 µg/L were able to diagnose ATI with a likelihood ratio of 5.97.34 NGAL is released from renal tubules in response to injury, rises before creatinine, and is not affected by volume depletion or diuretics.35–38 In animal studies, NGAL, along with liver fatty acid binding protein (LFABP), was shown to rise only modestly in volume depletion without histologic damage, as compared with the model of ischemia reperfusion.39 Lastly, kidney injury molecule–1 (KIM-1) was extensively expressed in biopsy specimens of patients with ATI, and urine concentrations of KIM-1 distinguished ATI from other causes of AKI.40 Once again, KIM-1 expression in renal tubules was dramatically increased after ischemia, demonstrating its biologic plausibility as a marker of true kidney injury.41 It has also been shown that urine NGAL improves the clinical model (consisting of multiple baseline characteristics including baseline eGFR, age, and history of diabetes) for predicting worsening AKI, defined as worsening AKIN stage and in-hospital death from an area under the curve (AUC) of 0.62–0.75, as compared with no improvement when adding FeNa to the model (AUC remained at 0.62).42 The use of IL-18, NGAL, LFABP, and KIM-1 in addition to the current paradigm would likely be able to distinguish between PRA and ATI, and ultimately help guide patient management.
In addition, the repurposing of existing clinical markers such as urine microscopy could be used to differentiate ATI from PRA. For example, the urine microscopy score, which is calculated from the number of granular casts and tubular cells, has been found to be higher in patients with ATI compared with those with PRA and was highly correlated with the urinary levels of IL-18, NGAL, and KIM-1. Patients with a microscopy score ≥3 (on the basis of the number of granular casts and renal tubular epithelial cells present in urine sample) had 3.5 times the risk of AKI progression defined as worsening AKIN stage and in-hospital mortality compared with those with microscopy scores of zero.42 Hence, in addition to novel biomarkers, clinically available biomarkers such as urine microscopy score could also be used to differentiate ATI from PRA.
Lastly, biomarker performance improves for diagnosis of AKI in the background of CKD. Urine KIM-1 levels diagnosed AKI more accurately in those with eGFR<60 ml/min per m2 as compared with those with normal baseline kidney function.43
Scenario 2: Differential Diagnosis in Cirrhosis
Clinical Dilemma
The diagnostic task of identifying the cause of AKI becomes even more challenging as a patient’s concomitant medical history becomes more extensive. AKI is a common complication in patients with advanced cirrhosis and is associated with significant mortality. In the setting of cirrhosis, the differential diagnosis broadens to include hepatorenal syndrome (HRS) along with PRA and ATI.44 The ability to make the distinction between HRS and ATI is critical, because treatments differ considerably. HRS may be reversed with restoration of renal perfusion, through vasoconstrictor therapy in addition to intravenous albumin treatment and subsequent liver transplantation. In contrast, patients with ATI should be treated with dialysis and considered for combined liver-kidney transplantation.45–48
Current Paradigm
The current criteria to diagnose HRS proposed by the International Club of Ascites (ICA) are presence of cirrhosis and ascites, serum creatinine >1.5 mg/dl without improvement after diuretic withdrawal and restoration of oncotic volume within 48 hours, absence of shock, lack of nephrotoxic agents, and absence of proteinuria, hematuria, and abnormal renal ultrasound.49 Notably, the ICA criteria rely heavily on the use of serum creatinine, but creatinine is especially unsuitable as a marker of kidney injury in the setting of cirrhosis because of low protein intake, loss of muscle mass, and decrease in synthesis secondary to hepatic injury.50 This will lead to false negative serum creatinine values and ultimately delay initiation of therapy. Many of the standard diagnostic tools, such as urine microscopy and FeNa<1%, have also traditionally been ineffective in this situation.51
Proposed Paradigm
In an effort to overcome the diagnostic limitations of serum creatinine, several biomarkers have been investigated in the setting of cirrhosis. Urine NGAL and IL-18 were able to discriminate between ATI and other causes of kidney impairment in patients with cirrhosis.52 The highest levels were found in patients with ATI, followed by HRS and PRA, respectively.46,53 The combination of several biomarkers was also assessed in diagnosing the type of AKI in patients with cirrhosis.54 Using a cut-off value of 365 ng/ml for NGAL, 85 pg/ml for IL-18, and 25 ng/ml for LFABP, the diagnosis of ATI in patients with cirrhosis was evaluated. Patients with cirrhosis who had at least one biomarker above the cut-off level were five times as likely to have ATI and those with all markers positive were 13 times as likely to have ATI. Biomarker levels were also able to increase the post-test probability of having ATI; for example, having an NGAL level >365 ng/ml raised the pretest probability from 40% to a post-test probability of 76%.54
Fortunately, several clinically available tests have also demonstrated an ability to distinguish between HRS and ATI. FeNa has long been marginalized in the setting of cirrhosis as the clinical cut-off of <1% is universal in these patients. However, its performance has recently been re-evaluated by multiple investigators.51,55,56 Although these studies have utilized different methods of diagnosing HRS, in all of their results, FeNa was significantly lower in cases of HRS than in cases of ATI. The median FeNa in patients with HRS across studies is remarkably uniform, averaging around 0.1%. In the study by Belcher et al., FeNa was lowest in patients with HRS (0.10% [interquartile range (IQR), 0.02–0.23]), and differed significantly from both levels in patients with PRA (0.27% [IQR, 0.13–0.58]), and levels in patients with ATI (0.31% [IQR, 0.12–0.65]).54 In the same study, spot albuminuria was investigated as an injury marker for ATI and was significantly higher in patients adjudicated with ATI (92 mg/dl [IQR, 44–254]) than it was in either PRA (21 mg/dl [IQR, 4–70]) or HRS (24 mg/dl [IQR, 13–129], P<0.001). The AUC identifying ATI at a cut-off of 44 mg/dl was 0.73 (IQR, 0.64–0.83).54 Until additional markers become clinically available, clinicians should consider incorporating FeNa and proteinuria into their diagnostic and prognostic approaches to joint AKI-cirrhosis cases (Figure 2). Although using this proposed paradigm may help discriminate between ATI and HRS, distinguishing PRA from HRS may be less reliable.
Guided clinical management in patients with cirrhosis by phenotyping AKI via biomarkers. AKI in patients with cirrhosis with a broad differential including HRS, ATI, and PRA. The use of novel biomarkers along side clinical markers in joint AKI-cirrhosis cases may distinguish between these three etiologies, as well as help guide clinical management.
Scenario 3: Management Decisions in Cardiorenal Syndrome
Clinical Dilemma
There are further clinical challenges associated with patients with cardiorenal syndrome. Currently, diuretics remain the mainstay of treatment in acute decompensated heart failure.57 However, there exists a frequently vexing clinical scenario of continuing with diuretic use, ignoring a rise in creatinine to attain optimal decongestion and effective perfusion pressure, or decreasing diuretic dose to avoid worsening renal function with diuresis.
Current Paradigm
A creatinine rise of >0.3 mg/dl in the setting of decompensated heart failure has been associated with longer hospital stay, higher readmission rates, and mortality.58,59 These elevations in creatinine and BUN, commonly known as “worsening renal function” in the cardiology literature, are associated with aggressive diuresis that is required to restore cardiovascular hemodynamics.60,61 Unfortunately, both creatinine and possibly high diuretic doses are markers rather than drivers of poor outcomes. There is a tremendous lack of evidence to guide this clinical dilemma mainly because the clinical markers such as BUN, creatinine, and urine output do not measure true renal injury.62 This clinical scenario has been further complicated by recent data suggesting significantly lower long-term mortality in patients developing hemoconcentration secondary to more aggressive diuretic dosing, despite a rise in creatinine.63–65
Proposed Paradigm
It has been shown that in patients with chronic heart failure urine NGAL is elevated, indicating a possibility of tubular damage.66 Similarly, urinary KIM-1 and N-acetyl-β-d-glucosaminidase were elevated in patients with chronic heart failure as compared with healthy controls despite a mean eGFR of 78 ml/min per m2, indicating the presence of tubular injury beyond that captured by creatinine.67 Therefore, low levels of NGAL, N-acetyl-β-d-glucosaminidase, and KIM-1 might aid in the decision to continue with intense diuretic regimens despite a rise in creatinine. These biomarkers could potentially be evaluated in future studies or in completed trials by examining banked samples such as those from the Diuretic Optimization Strategies Evaluation trial to evaluate renal injury in the setting of different intensities of diuresis.68
Clinically established markers such as urinary excretion of albumin (UAE) may reliably guide diuretic management in patients with decompensated heart failure.69 UAE as defined by 1+ proteinuria on dipstick was associated with greater decline in eGFR over 23 months in patients with moderate-to-severe heart failure as compared with patients with negative protein on dipstick.69 In addition to UAE, the Evaluation Study of Congestive Heart Failure and Pulmonary Artery Catheterization Effectiveness trial showed that markers of hemoconcentration were associated with worsening renal function (defined by decline in eGFR), but nonetheless lower mortality.63 Hemoconcentration was defined as an increase in at least two of the following: serum albumin, serum total protein, and hematocrit. Therefore, the application of existing clinical markers such as UAE, hematocrit, serum albumin, and serum protein, along with novel biomarkers, might generate a panel to distinguish between causes leading to creatinine rise with the potential for therapeutic diuretic success in decompensated heart failure.
Scenario 4: Allocation of Kidneys from Deceased Donors
Clinical Dilemma
Over 100,000 patients are awaiting kidney transplants in the United States and unfortunately the number of deceased donor transplants has plateaued around 11,000. This necessitates efforts to salvage all viable procured kidneys.70 Nevertheless, about 2500 procured kidneys from deceased donors are discarded each year. AKI kidneys are discarded at much higher rates secondary to risk of delayed graft function (DGF), risk of increased hospital stay post-transplantation, and concerns of long-term graft failure.71 However, it has been shown that donor AKI, especially in patients who later develop DGF, does not preclude favorable recipient allograft outcomes, and may in fact portend improved 6-month eGFR with increasing AKI severity using AKIN staging.72
Current Paradigm
To help address these shortcomings via better organ utilization, the national kidney allocation system was revised in December of 2014 to incorporate the kidney donor profile index.73 The kidney donor profile index comprises ten donor characteristics, which include terminal serum creatinine as well as donor age, height, weight, ethnicity, history of hypertension, history of diabetes, cause of death, hepatitis C status, and donation after circulatory death. These ten factors generate a score that is presented as percentage risk of graft failure compared with kidneys retrieved in the prior year. However, this panel has its limitations. For example, the use of terminal serum creatinine is especially restrictive around the time of death given that it is affected by multiple nonrenal factors such as altered creatinine kinetics, muscle injury, and administration of fluids, and a single value of creatinine does not fully account for the acute trajectory or chronicity of renal function. Wedge biopsy specimens of donor kidneys obtained at time of procurement are also not accurate, can overestimate the amount of glomerulosclerosis, and are suboptimal for the diagnosis of ATI.74 Consequently, there remains a compelling need for more reliable tools to assess donor kidney quality in the presence of AKI.
Proposed Paradigm
To delineate which donor AKI kidneys will have poor versus favorable outcomes in recipients, both kidney injury and repair biomarkers have been evaluated in deceased donor urine. Two injury biomarkers, NGAL and LFABP, provided modest incremental value in predicting worse recipient 6-month eGFR, especially in those without DGF.75 On the contrary, a repair phase protein known as YKL-40 was associated with less DGF and better long-term eGFR.76
Preclinical studies on YKL-40 have generated strong support for its cytoprotective and reparative roles in the setting of kidney injury.77 Among those donor kidneys with clinically defined injury before transplantation, the ones with the greatest urinary YKL-40 levels have the best chance of successfully regaining function.76 The use of a recovery marker like YKL-40 in assessing deceased donor kidneys has immense potential, because it could increase the number of transplanted AKI kidneys by feasibly decreasing discard rates among AKI kidneys and providing more wait-listed patients with the opportunity to receive a viable kidney.
Discussion
Use of Biomarkers in Clinical Management
Although novel biomarkers can address diagnostic delay in AKI and differentiate between PRA and ATI, further research is needed to advance biomarkers to bedside. As a strategy, a recent clinical trial used plasma NGAL for enrollment of participants in the assessment of early versus late initiation of renal replacement therapy for AKI.78 In addition, the FDA approved the first AKI point-of-care biomarker device, NephroCheck.79 It measures urinary levels of tissue inhibitor of metalloproteinase-2 and IGF-binding protein 7, which are cell cycle arrest biomarkers. Using NephroCheck, the positive predictive value to diagnose stage 2 or 3 AKI as defined by KDIGO is 49% and the negative predictive value is 97%.80–82 The effect of NephroCheck will be determined once it is adopted into clinical practice, but it does offer the ability to identify risk of imminent AKI in critically ill patients and potentially identify high-risk patients for enrollment in clinical trials. However, as emphasized by the ADQI, perhaps the combination of biomarkers of kidney damage and kidney function, rather than a biomarker representation of a single biologic process such as cell cycle arrest, will be needed to facilitate the diagnosis of AKI.83 This concept was validated in 345 children undergoing cardiopulmonary bypass, where the combination of functional (cystatin C) and damage (NGAL) biomarkers was superior than change in serum creatinine in predicting severity and persistence of AKI.84
More so, the use of biomarkers without proper clinical risk stratification will yield to suboptimal biomarker performance. To overcome this challenge, the renal angina index was established to risk-stratify patients with AKI on the basis of the severity of the clinical setting and the percentage change in creatinine clearance. This strategy was validated in a pediatric population where the incorporation of biomarkers into the renal angina index improved net reclassification of predicting AKI.85
Conclusion and Future Direction
There may be successful biomarkers among the current candidates, but their full potential cannot be realized when they are compared with the confining indicator that is serum creatinine. It is also possible that AKI is a syndrome with each setting having a unique set of case-mix and pathophysiology with a distinctive set of clinical decisions, necessitating an individualized panel of biomarkers for each setting. In an effort to move away from the sole use of creatinine in diagnosing renal injury, the National Institutes of Health has initiated the Kidney Precision Medicine Project which is a mission toward obtaining and assessing kidney biopsy specimens to create a kidney tissue atlas and a movement toward tissue-driven definitions of renal injury and disease.86 After this hopeful transformation from creatinine-defined to tissue-defined renal injury, the field of nephrology will likely develop a deeper understanding of kidney injury biomarkers, which could lead physicians to incorporate them routinely into clinical practice and better personalize patient care.
Disclosures
None.
Acknowledgments
C.R.P. is supported by National Institutes of Health (NIH) grants RO1HL085757 and K24DK090203. S.G.M. is supported by the T32 training grant (T32DK007276) from the NIH.
Footnotes
Published online ahead of print. Publication date available at www.jasn.org.
- Copyright © 2017 by the American Society of Nephrology