Phase 1: Discovery of Potential Biomarkers through Unbiased or Hypothesis-Generating Exploratory Studies

The primary goal of phase 1 is to identify potential leads using various technologies and to confirm and prioritize the identified leads. The search for biomarkers often begins with preclinical studies that compare either tissue or biologic fluids in diseased animals (e.g., animals with kidney injury) with those in healthy animals to identify genes or proteins that appear to be upregulated or downregulated in diseased tissue relative to control tissue. When biologic samples, such as blood and urine, are readily available from humans, it is possible to forgo the animal model stage. Innovative discovery technologies include microarray-based gene expression profiling that provides information regarding expression of genes, microRNA-based expression, and proteomic as well as metabolomic profiling of biologic fluids based on mass spectrophotometry and other technologies. The candidate marker approach, especially when informed by the pathophysiology of the disease for which the biomarker is being evaluated, should not be ignored.

Once a promising biomarker is discovered, the validation process begins. An assay has to be developed and validated. The validation process is laborious and expensive, requiring access to patient samples with complete clinical annotation and long-term follow-up, as described in the discussion of phase 2. In addition, each biomarker must be qualified for specific application. This statement is especially true in the case of kidney diseases, for which one biomarker alone may not satisfy the requirements of an ideal biomarker. This situation is described in the subsequent section on phase 4. Incorporation of several of these novel biomarkers into a biomarker panel may enable simultaneous assessment of site-specific kidney injury or several mechanisms contributing to clinical syndromes.

Phase 2: Development and Validation of an Assay for the Measurement or Identification of the Biomarker in Clinical Samples

The primary goal of phase 2 is to develop and validate a clinically useful assay for a biomarker that has the ability to distinguish a person with kidney disease/injury from persons with healthy kidneys in a high-throughput fashion. This phase involves development of an assay, optimization of assay performance, and evaluation of the reproducibility of the assay results within and among laboratories. Defining reference ranges of biomarker values is a crucial step before the biomarker can be used clinically. It is important to characterize how the levels of these markers vary with patient age, sex, and race or ethnicity, and how biomarker values are related to known risk factors.

Phase 3: Demonstration of the Biomarker’s Potential Clinical Utility in Retrospective Studies

In phase 3 of biomarker development, the primary objectives are to (1) evaluate the biomarker potential in samples obtained from a completed clinical study, (2) test the diagnostic potential of the biomarker for early detection, and (3) determine the sensitivity and specificity of the biomarker using defined threshold values of the biomarker for utility in prospective studies. For instance, if the levels of biomarker differ significantly between subjects with acute or chronic kidney injury and control subjects only at the time of clinical diagnosis, then the biomarker shows little promise for population screening or early detection. In contrast, if levels differ significantly at hours, days, or years before clinical symptoms appear, then the biomarker’s potential for early detection is increased. This phase also involves comparing the biomarker with several other novel biomarkers or existing “gold standard” biomarkers and defining the biomarkers’ performance characteristics (i.e., sensitivity, specificity) using receiver operating characteristic curve analysis. This latter process is particularly challenging in kidney disease, given uncertainties in the sensitivity and specificity of the gold standard used.

Phase 4: Performance of Prospective Screening Studies

The primary aim of phase 4 studies is to determine the operating characteristics of the biomarker in a relevant population by measuring detection rate and false referral rate. In contrast to phase 1, 2, and 3 studies, which are based primarily on stored specimens, studies in phase 4 involve screening subjects prospectively and demonstrating that clinical care is changed as a result of the information provided by the biomarker analysis.

Phase 5: Continued Assessment of the Validity of the Biomarker in Routine Clinical Practice

Phase 5 addresses whether measurement of the biomarker alters physician decision making and/or reduces mortality or morbidity associated with the given disease in the population.

Serum Glomerular Filtration Markers

During the course of injury, kidney function may be impaired with reduction in the GFR and accumulation of several nitrogenous compounds in the blood. Serum creatinine and BUN concentrations are routinely used as markers of kidney injury, but it is important to recognize these parameters as markers of kidney dysfunction rather than as direct markers of injury.

As discussed elsewhere in this text, the estimated GFR (eGFR), using creatinine as a biomarker, is most reliable for CKD under steady-state conditions. In the acute setting, its use is more problematic for reasons that have already been discussed. In healthy persons, the GFR is in the range of 90 to 130 mL/min/1.73 m ² . By definition, patients with stage 4 or 5 CKD have GFRs that are below 30 mL/min/1.73 m ² . Complications of CKD are more pronounced at lower GFRs, and mild to moderate CKD may progress to ESKD.

In AKI, the GFR is only indirectly linked to kidney injury, and changes in the GFR reflect a late consequence in a sequence of events associated with a primary insult to the kidney. Furthermore, because of renal reserve, a large amount of functioning renal tissue can be lost without significant changes in the GFR. The functional effects of renal reserve on the GFR can be demonstrated in kidney donors, who often have only modest changes in serum creatinine levels and the GFR after donating one kidney, even though half of the renal mass is lost.

Ideally, a serum GFR marker should be freely filtered with no reabsorption or secretion in the kidney and should maintain a constant plasma level when kidney function is stable. GFR can be determined using exogenous and endogenous markers of filtration. Evaluation of the GFR using the exogenous marker inulin, iothalamate, or iohexol provides reliable results and represents the gold standard; however, the process is time consuming and expensive and can be performed only in specialized settings. Once the GFR level falls below 60 mL/min/1.73 m ² , renal functional impairment can be estimated adequately from the serum creatinine level, various equations being used to calculate the eGFR. Although traditionally, these equations have been less accurate for patients with higher GFRs, newer formulas have been constructed utilizing more patients with normal and near normal GFRs.

Urinary Glomerular Cell Injury Markers

Defects in podocyte structure have been reported in many glomerular diseases, which have been classified as “podocytopathies.” Injured podocytes have been reported in immunologic and nonimmunologic forms of human glomerular disease, including hemodynamic injury, protein overload states, injury from environmental toxins, minimal change disease, focal segmental glomerulosclerosis, membranous glomerulopathy, diabetic nephropathy, and lupus nephritis. Podocytes may be injured in many forms of human and experimental primary glomerular disease and in secondary forms of focal segmental glomerulosclerosis, including that caused by hypertension, diabetes, and tubulointerstitial disease. Before detachment from the glomerular basement membrane, podocytes undergo structural changes, including effacement of foot processes and microvillous transformation.

Urine Microscopy

Urine microscopy with sediment examination is a time-honored test that is routinely used to assist in the diagnosis of kidney injury. The urine from patients with tubular injury typically contains proximal tubular epithelial cells, proximal tubule epithelial cell casts, granular casts, and mixed cellular casts. Patients with predominantly prerenal azotemia occasionally have hyaline or fine granular casts in their urine. Several studies have shown that the increase in urinary cast excretion correlates well with AKI. Marcussen and associates demonstrated that patients with tubular injury had a high number of granular casts than those with prerenal azotemia.

There has now been a resurgence in urinalysis sediment scoring systems for the diagnosis of AKI. Several of these systems have shown excellent specificity for AKI and correlate well with severity of AKI. However, their widespread acceptance has been hampered by the relatively modest sensitivity of urine microscopy for detecting AKI. Urine microscopy remains a user-dependent tool that displays a tremendous amount of interphysician variability, a feature that likely contributes to its suboptimal sensitivity for AKI. Three of the most widely reported urine microscopy scoring systems are reviewed in Table 30.5 .

Several later studies have looked at the potential of using urine microscopy in combination with other biomarkers for tubular injury with varying degrees of success. In the near future, urine microscopy, a current mainstay in the clinical diagnosis of AKI, could be used in concert with markers of glomerular function and validated biomarkers of tubular injury to diagnose AKI.

α ₁ -Microglobulin

α-Microglobulin is a low-molecular-weight glycoprotein of approximately 27 to 30 kDa and a member of the lipocalin superfamily. It is primarily synthesized by the liver and is available both in free form and as a complex with IgA. α ₁ -Microglobulin has been detected in human serum, urine, and cerebrospinal fluid. Urine and serum values have been found to be elevated in patients with renal tubular diseases. α ₁ -Microglobulin is freely filtered at the glomerulus and completely reabsorbed and catabolized by the normal proximal tubule. Megalin mediates the uptake of this protein in the proximal tubule. Therefore, an increase in the urinary concentration of α ₁ -microglobulin indicates proximal tubular injury or dysfunction. The urinary levels of α ₁ -microglobulin are influenced by age. The normal range in populations younger than 50 years is less than 13 mg per g of creatinine and in those 50 years or older is less than 20 mg per g of creatinine. In comparison with β ₂ -microglobulin, α ₁ -microglobulin is more stable over a range of pH levels in the urine, making it a more acceptable urinary biomarker.

β ₂ -Microglobulin

β ₂ -Microglobulin is a low-molecular-weight polypeptide with a molecular weight of 11.8 kDa. It is present on the cell surfaces of all nucleated cells and in most biologic fluids, including serum, urine, and synovial fluid. β ₂ -Microglobulin is normally excreted by glomerular filtration, reabsorbed almost completely (approximately 99%), and catabolized by the normal proximal tubule in humans. Megalin mediates the uptake of this protein in the proximal tubule. In healthy individuals, approximately 150 to 200 mg of β ₂ -microglobulin is synthesized daily with a normal serum concentration of 1.5 to 3 mg/L. Any pathologic state that affects kidney function results in an increase in β ₂ -microglobulin levels in the urine because of the impeded uptake of β ₂ -microglobulin by renal tubular cells. For spot urine collections, the concentration of β ₂ -microglobulin in healthy individuals is typically 160 µg/L or less or 300 µg per g of creatinine or less. Unlike serum levels of urea, those of β ₂ -microglobulin are not influenced by food intake, making this polypeptide an attractive marker for malnourished patients with low serum urea levels. In patients with CKD, increases in serum β ₂ -microglobulin levels reflect the decrease in glomerular function. In patients with ESKD, serum levels of β ₂ -microglobulin are usually in the range of 20 to 50 mg/L. β ₂ -Microglobulin accumulation is linked to toxicity because the molecule precipitates and forms fibrillary structures and amyloid deposits, particularly in bone and periarticular tissue, leading to the development of carpal tunnel syndrome and erosive arthritis. Elevations of β ₂ -microglobulin have been reported in several AKI and CKD clinical settings, including in cadmium toxicity and following cardiac surgery, liver transplantation, and renal transplantation. In idiopathic membranous nephropathy, β ₂ -microglobulin level was identified as a superior independent predictor of the development of renal insufficiency. Other studies have reported that β ₂ -microglobulin performs as well as, if not better than, serum creatinine for the detection of acute kidney injury in critically ill children or after cardiac surgery in adults.

Serum concentrations of β ₂ -microglobulin should be interpreted cautiously because they are altered significantly in various diseases, including rheumatoid disorders and several types of cancers. Initially, it was believed that the increase in β ₂ -microglobulin levels in CKD is solely due to declines in kidney function, but later studies have shown that other factors, including increased synthesis of β ₂ -microglobulin, may contribute in patients with ESKD. Another significant drawback associated with the use of urinary β ₂ -microglobulin as a marker of kidney injury is its instability in urine at room temperature, particularly when the pH is less than 5.5; for this reason, the urine should be alkalinized and frozen at −80° C immediately after collection.

Glutathione S-Transferase

Primarily two subtypes of the enzyme glutathione S-transferase (GST) are found in the kidney. α-GST is found mainly in the proximal tubular cells, whereas π-GST is found predominantly in the distal tubular epithelial cells. Elevation of urinary α-GST has been reported in several animal models treated with nephrotoxic drugs or after ischemic renal injury. However, in a prospective study of patients with sepsis admitted to the ICU, α-GST levels were no different in patients in whom AKI developed and in patients without AKI. π-GST levels were higher in all patients with sepsis than in healthy volunteers but π-GST levels were not predictive of AKI as defined by the AKIN criteria. In one prospective study, the value of tubular enzyme levels in predicting AKI was assessed in 26 critically ill adult patients admitted to the ICU. AKI developed in 4 patients, and ROC analysis showed that γ-glutamyl transpeptidase, π-GST, α-GST, alkaline phosphatase, and NAG had excellent discriminating power for AKI (AUCs = 0.950, 0.929, 0.893, 0.863, and 0.845, respectively). Both α-GST and π-GST have demonstrated limited ability to detect AKI following adult cardiac surgery. However, in a small single-center study of 123 subjects, π-GST did demonstrate the ability to detect which patients with AKIN stage 1 AKI would go on to progress to stage 3 or need RRT (AUC = 0.86; P = 0.002); however, to date, these data have not been validated in a larger cohort.

In kidney transplant recipients, increased levels of α-GST were associated with cyclosporine A toxicity, whereas π-GST elevation was associated with acute allograft rejection. In a cross-sectional study of patients with diabetes, the relationships between urine albumin/creatinine ratio and urinary levels of collagen IV, α-GST, and π-GST were assessed. Levels of all three markers were directly (albeit weakly) correlated with urine albumin/creatinine ratio, but a progressive increase in the proportion of patients with abnormal biomarker levels in those with normal urine albumin levels, microalbuminuria, and macroalbuminuria was observed only for collagen IV and π-GST.

Hepcidin-25

Hepcidin-25, a 2.8-kDa hormonal regulator of iron metabolism, is produced in the liver, heart, and kidney. Hepcidin binds and induces the internalization and degradation of the transmembrane iron exporter ferroportin. Hepcidin acts to downregulate iron uptake and reduce extracellular iron availability from stored iron. Given its link to iron metabolism and the fact that free iron is known to be released in the setting of the ischemia reperfusion injury and oxidative stress that occurs with cardiopulmonary bypass, urinary hepcidin-25 has been investigated as a marker of kidney injury following cardiac surgery. Ho and colleagues identified urinary hepcidin-25 in a nested case-control study of 44 adults who underwent cardiac surgery. Using surface-enhanced laser desorption/ionization time-of-flight mass spectrometry (SELDI-TOF-MS) on urine samples from 22 individuals in whom at least RIFLE Risk category AKI developed and 22 individuals whose creatinine did not increase more than 10% from baseline during the postoperative period (no AKI), these researchers found that hepcidin-25 was dramatically upregulated in the urine of patients with no AKI. Taking this a step further, the same group quantified the concentration of hepcidin in the urine (normalized to urine creatinine) and demonstrated that concentrations were higher in those who did not go on to have postoperative AKI ( P < 0.0005). In a multivariate analysis, hepcidin-25 was significantly associated with the avoidance of AKI with urinary concentrations on postoperative day 1, providing an AUC of 0.80. The data from this small study have been corroborated by those from another modest-sized cohort of 100 adults undergoing cardiopulmonary bypass. Haase-Fielitz and associates demonstrated that 6 hours after cardiopulmonary bypass, urinary hepcidin-25 levels were lower in the nine subjects in whom RIFLE-based AKI developed than in those with no AKI (AUC = 0.80; P = 0.004). These results warrant further preliminary investigations in other AKI settings as well as prompt validation in a larger cohort of cardiac surgery patients.

Interleukin-18

IL-18 is an 18-kDa proinflammatory cytokine that is activated by caspase-1 and is produced by renal tubular cells and macrophages. Animal studies indicate that IL-18 is a mediator of acute tubular injury, including both neutrophil and monocyte infiltration of the renal parenchyma. Studies have shown that caspase-1 knockout mice experienced the same degree of ischemic AKI as wild-type mice injected with an IL-18–neutralizing antiserum, demonstrating that IL-18 is an important mediator of ischemic AKI. Others have shown that IL-18 plays a major role in macrophage activation, with mice engrafted with IL-18–deficient bone marrow experiencing less AKI than those with IL-18–replete marrow. Similarly, in IL-18 knockout mice with AKI, tumor necrosis factor-α, inducible nitric oxide synthase, macrophage inflammatory protein-2, and monocyte chemoattractant protein-1 mRNA expression are all decreased, speaking to the deleterious impact of IL-18 in AKI. In the human kidney, IL-18 is induced and cleaved mainly in the proximal tubules and released into the urine. IL-18 has been shown to participate in a variety of renal disease processes, including ischemia-reperfusion injury, allograft rejection, infection, autoimmune conditions, and malignancy. IL-18 is easily and reliably measured in urine by commercially available ELISA and microbead-based assays.

Acute Kidney Injury

Several studies have demonstrated the usefulness of IL-18 as a biomarker for detection of AKI. Originally, Parikh and associates studied a group of 72 patients and reported urinary IL-18 levels significantly higher in patients diagnosed with acute tubular necrosis (ATN) than in patients with prerenal azotemia or urinary infection and in healthy control subjects with normal renal function. Since then, several large multicenter studies have gone on to investigate the ability of IL-18 to detect AKI in a variety of clinical settings.

The TRIBE-AKI Consortium measured IL-18 in 1219 adults who underwent cardiac surgery. Identification of those at high risk for postoperative AKI required the presence of one of the following: (1) emergency surgery, (2) preoperative serum creatinine higher than 2 mg/dL, (3) left ventricular ejection fraction < 35%, (4) New York Heart Association stage III or IV heart failure—left ventricular function, (5) age more than 70 years, (6) preexisting diabetes mellitus, (7) and concomitant coronary artery bypass grafting (CABG) and valve surgery or (8) repeat cardiac surgery. Those with preoperative AKI, kidney transplants, ESKD, or a preoperative serum creatinine higher than 4.5 mg/dL were excluded. After dividing the cohort into quintiles according to IL-8 level, compared to the lowest quintile, the highest quintile of IL-18 was associated with a 6.8-fold higher risk of AKI, defined as a postoperative doubling of serum creatinine or requirement for acute dialysis. The first postoperative concentration of IL-18 (0-6 hours) provided an AUC of 0.74, which increased to 0.76 after IL-18 values were combined with a clinical model of factors known to impact AKI risk. The TRIBE-AKI pediatric cohort (311 children) reported results in line with the TRIBE adult study, in which the highest quintile of IL-8 values was associated with 9.4-fold increased risk of AKI (doubling of serum creatinine or dialysis) in comparison with the lowest quintile. The effect was slightly attenuated after adjustments were made for clinical factors known to impact AKI (adjusted OR, 28.8; 95% CI, 6.9 (1.7 to 28.8).

In a secondary analysis of the adult cohort, Koyner and associates demonstrated that IL-18 value at the time of AKI can identify those with early AKI (AKIN stage 1) who will progress to more severe stages of AKI (AKIN stage 2 or 3). Of the 380 adults in whom at least stage 1 AKI developed, 45 went on to have stage 2 or 3 disease. In the entire cohort, those whose IL-8 values were in the fifth quintile were at increased risk for development of progressive AKI (OR, 3.63; 95% CI, 1.64 to 8.03) and this effect was only slightly attenuated after adjusting for the clinical model (OR, 3.00; 95% CI, 1.25 to 7.25).

Finally, when investigated in a separate secondary analysis of this cohort, IL-18 concentrations collected in the immediate postoperative period were associated with long-term mortality following cardiac surgery. This later investigation provided a median follow-up of 3.0 years (interquartile range [IQR], 2.2 to 3.6), during which 139 of the 1199 subjects died (50 deaths per 1000 person-years). After adjustments were made for clinical factors known to affect mortality, in patients without AKI (n = 792), those whose IL-18 values were in the third tertile were at increased risk of long-term mortality in comparison with those whose values were in the first tertile (adjusted HR, 1.23; 95% CI, 1.02 to 1.48). This effect was magnified in those subjects with perioperative AKI (n = 407), in whom those with values in the third tertile had an adjusted hazard ratio of 3.16 (95% CI, 1.53 to 6.53) in comparison with the reference cohort. Thus IL-18 provides additional prognostic information about long-term postoperative mortality in patients with and without AKI.

When investigated in the setting of critical illness and ICU admission, IL-18 has not demonstrated the same robust results. In a study of 451 critically ill subjects, in 86 of whom AKI developed within the first 48 hours, Siew and colleagues demonstrated that urine IL-18 did not reliably predict AKI. Although IL-18 levels at the time of ICU admission were higher in those who went on to have AKI, the AUC was 0.62 (95% CI, 0.54 to 0.69) and only marginally improved to 0.67 after the exclusion of those with prior known CKD (eGFR > 75 mL/min). Despite the inability to reliably detect AKI, urine IL-18 levels did correlate with other adverse patient outcomes, including the need for RRT and 28-day mortality. This poor performance in the setting of critical illness has been corroborated by other studies, including a post hoc analysis of data from the EARLYARF trial. In this prospective observational study in two large general ICUs (n = 529), IL-18 provided an AUC of only 0.62 for the diagnosis of AKIN stage 1 AKI but once again performed much better at forecasting the need for RRT or death (within 7 days). Unlike other biomarker studies, this IL-18 study did not demonstrate improved predictive powers with the cohort stratified according to pre-admission CKD stage. In a separate post hoc analysis of the same EARLYARF cohort, urinary IL-18 concentrations were shown to be significantly higher in patients with prerenal azotemia (defined as AKI that recovered within 48 hours of ICU admission and was associated with a fractional excretion of sodium < 1%; n = 61) than in those with no AKI (n = 285). There was a trend toward higher values in those with AKI (n = 114, non-prerenal) than in those with prerenal AKI ( P = 0.053).

In a single-center study of 339 mixed surgical and medical ICU patients, Doi and colleagues demonstrated that IL-18 values were significantly elevated in those with both established and newly diagnosed AKI at the time of ICU arrival. Although biomarker concentrations and AUCs were higher in those with established AKI (AUC = 0.78) than in those with newly diagnosed AKI (AUC = 0.59), concentrations and AUCs in both of these subgroups were significantly different from those in the “no AKI” cohort. In this same study, IL-18 levels were significantly higher in nonsurvivors. These results were fairly similar to those reported by Nikolas and colleagues, who measured urinary biomarkers in 1635 ED patients at the time of admission and compared the values with adjudicated AKI outcomes in which prerenal AKI was defined as RIFLE Risk category that returned to baseline within 72 hours and was in the clinical setting, suggesting decreased transient effective circulating volume. These researchers demonstrated that IL-18 values for patients with more severe intrinsic AKI were significantly higher than those for patients with prerenal AKI. However, there was no difference in values between those with prerenal AKI and those with no AKI.

When measured at kidney transplantation, IL-18 level accurately predicted delayed graft function (AUC = 0.90) and predicted the rate of decline in serum creatinine concentration. In patients with diabetic kidney disease and proteinuria, IL-18 levels in renal tubular cells are higher than in patients with nondiabetic proteinuric disease. To understand the utility of IL-18 and urinary NGAL in predicting graft recovery after kidney transplantation, Hall and colleagues conducted a prospective, multicenter, observational cohort study of recipients of deceased-donor kidney transplants. They collected serial urine samples from 91 patients for 3 days after transplantation. After adjustment for recipient and donor age, cold ischemia time, urine output, and serum creatinine concentration, NGAL and IL-18 concentrations accurately predicted the need for dialysis in transplant recipients. Furthermore, NGAL and IL-18 concentrations predicted graft recovery up to 3 months later. In further follow-up of this cohort, urine IL-18 concentrations collected at the time of surgery were correlated with graft outcomes 1 year after transplantation. Upper median values of IL-18 on the first postoperative day had an adjusted OR of 5.5 (95% CI, 1.4 to 21.5) and poor graft function, defined as a GFR < 30 mL/min or return to RRT.

In a study by Ling and associates involving patients who underwent coronary angiography, urinary IL-18 and NGAL concentrations were significantly increased at 24 hours after the procedure in those in whom contrast medium–induced nephropathy developed but not in the control group. ROC curve analysis demonstrated that both IL-18 and NGAL showed better performance in early diagnosis of contrast nephropathy than serum creatinine ( P < 0.05). Importantly, elevated urinary IL-18 concentrations 24 hours after contrast administration were also found to be an independent predictive marker for later major cardiac events (relative risk [RR] = 2.1).

Kidney Injury Molecule-1

Kidney injury molecule-1 (KIM-1 in humans, Kim-1 in rodents), which is also referred as T cell immunoglobulin and mucin domains–containing protein-1 ( TIM-1 ) and hepatitis A virus cellular receptor-1 ( HAVCR-1 ), is a type I transmembrane glycoprotein with an ectodomain containing a six-cysteine immunoglobulin-like domain, two N -glycosylation sites, and a mucin domain. In an effort to identify molecules involved in kidney injury, Bonventre’s group originally discovered Kim-1 using representational difference analysis (a PCR-based technique) in rat models of acute ischemic kidney injury. Importantly, KIM-1 was shown to be significantly expressed in kidneys, specifically in proximal tubular cells of humans after ischemic injury, whereas it was virtually absent or present at low levels in healthy kidneys. KIM-1 has evolved as a marker of proximal tubular injury, the hallmark of virtually all proteinuric, toxic, and ischemic renal diseases. KIM-1 has been shown to be a highly sensitive and specific marker of kidney injury in several rodent models, including models of injury due to ischemia, cisplatin, folic acid, gentamicin, mercury, chromium, cadmium, contrast agents, cyclosporine, ochratoxin A, aristolochic acid, d -serine, and protein overload.

In 2002, the Bonventre group published the first clinical study linking urinary levels of KIM-1 with AKI, demonstrating that tissue expression of KIM-1 is correlated with the severity of acute tubular necrosis and corresponding levels of KIM-1 ectodomain in the urine of patients with clinically significant AKI. Since then, numerous other studies have been published on the ability of KIM-1 to detect AKI in a variety of settings, including cardiac surgery, critical illness, and general hospitalized AKI, which has been mixed.

Later, KIM-1 was investigated in several multicenter larger trials. In the TRIBE-AKI adult and pediatric cardiac surgery cohorts, the fifth quintile of urinary KIM-1 values was associated with a risk of AKI (defined as postoperative doubling of serum creatinine concentration or need for acute dialysis) 6.2-fold greater than that for the lowest KIM-1 values. This risk remained significant (4.8-fold) after adjustment of data for the clinical model used (age, race, sex, cardiopulmonary bypass time, non-elective surgery, preoperative GFR, diabetes, hypertension, and study center). The effect was completely attenuated after urinary IL-18 and plasma and urine NGAL were included in the model. As for the long-term mortality in the TRIBE cohort, the third tertile of perioperative KIM-1 concentrations was found to be associated with increased mortality. In patients without AKI (n = 792), the third tertile had an adjusted HR of 1.83 (95% CI, 1.44 to 2.33), whereas in the 407 subjects with AKI, the adjusted HR was slightly higher, at 2.01(95% CI, 1.31 to 3.1). In the pediatric cohort, those with KIM-1 values in the fifth quintile were at increased risk of AKI in the unadjusted analysis; however, this effect failed to remain significant after a pediatric adjustment model was applied.

The data on KIM-1 in the setting of critical illness–related AKI have been just as mixed as the perioperative results, with KIM-1 providing an AUC of 0.66 (95% CI, 0.61 to 0.72) for the diagnosis of AKI in samples from the EARLYARF study. The results were less impressive with regard to the prediction of dialysis (AUC = 0.62) or death (AUC = 0.56) within the first week following ICU admission. In the cohort of 529 mixed ICU patients, KIM-1 outperformed other biomarkers in its ability to detect AKI at the time of ICU admission in those with a preadmission GFR < 60 mL/min (AUC, 0.7; 95% CI, 0.58 to 0.82). In a separate post hoc analysis of this cohort, KIM-1 levels demonstrated a significant stepwise increase in the comparison of those with no AKI (median serum creatinine 170 µg/mmol [IQR, 69 to 445]), those with prerenal AKI (median serum creatinine 291 µg/mmol [IQR, 121 to 549]), and those with intrinsic AKI (lasting more than 48 hours) (serum creatinine 376 µg/mmol [IQR, 169 to 943]). In addition, KIM-1 demonstrated the ability to forecast the development of intrinsic AKI at the time of ED arrival with an AUC of 0.71 (95% CI, 0.65 to 0.76; P < 0.001). KIM-1 values again increased in a stepwise fashion, values being lowest in those with no AKI or CKD and rising in other subjects in the following order: stable CKD, prerenal AKI, and intrinsic AKI. Additionally, KIM-1 values were able to forecast inpatient mortality and the need for RRT.

The usefulness of KIM-1 has been demonstrated not only as a urinary marker but also as a tool for evaluating kidney injury in kidney biopsy specimens by immunohistochemical methods. For example, Van Timmeren and associates found that the level of KIM-1 protein expression in proximal tubular cells correlated with tubulointerstitial fibrosis and inflammation in kidney tissue specimens from 102 patients who underwent kidney biopsy for a variety of kidney diseases. In a subset of patients whose urine was collected near the time of biopsy, urinary KIM-1 levels correlated with tissue KIM-1 expression in 100% of biopsy samples from patients with deterioration in kidney function and histologic changes indicative of tubular damage. In biopsy specimens from transplanted kidneys, greater KIM-1 staining was detected in 100% of patients with deterioration of kidney function and pathologic changes indicating tubular injury, in 92% of patients with acute cellular rejection, and in 28% of patients with normal biopsy findings. In contrast, Hall and associates demonstrated that urinary KIM-1 levels did not correlate with peritransplantation or 1-year graft function. Similarly, Schroppel and colleagues investigated KIM-1 RNA expression in perioperative samples collected from both living- and deceased-donor kidneys and found no significant correlation between KIM-1 staining and the occurrence of delayed graft function.

Liver-Type Fatty Acid–Binding Protein

Urinary fatty acid–binding protein 1 (FABP1) has been proposed to be a useful biomarker for early detection of AKI and monitoring of CKD. Also known as L-type or liver-type fatty acid–binding protein (L-FABP), which will be used in this book, FABP1 was first isolated in the liver as a binding protein for oleic acid and bilirubin. FABP1 binds selectively to free fatty acids and transports them to mitochondria or peroxisomes, where free fatty acids are β-oxidized and participate in intracellular fatty acid homeostasis. There are several different types of FABP, which are ubiquitously expressed in a variety of tissues. At this time, nine different FABPs have been reported: liver (L), intestinal (I), muscle and heart (H), epidermal (E), ileal (I1), myelin (M), adipocyte (A), brain (B), and testis (T). L-FABP is expressed in proximal tubules of the human kidney and localized in the cytoplasm. Increased cytosolic L-FABP in proximal tubular epithelial cells may derive not only from endogenous expression but also from circulating L-FABP that might be filtered at the glomeruli and reabsorbed by tubular cells.

Susantitaphong and associates published a meta-analysis reporting the performance of L-FABP from 15 prospective cohorts and two case-control studies. Although the researchers were able to meta-analyze only 7 of the cohort studies, they demonstrated that L-FABP levels were 74.5% sensitive (95% CI, 60.4% to 84.8%) and 77.6% specific (95% CI, 61.5% to 88.2%) for the diagnosis of AKI. Additionally, they demonstrated that the results were more promising for the predication of in-hospital mortality. They concluded that on the basis of the low quality of many of the studies and the varied clinical settings, L-FABP may be a promising biomarker for the early detection of AKI. In this discussion we highlight some of the larger and later clinical investigations of L-FABP.

Portilla and colleagues demonstrated that L-FABP predicts the development of AKI within 4 hours of surgery in children undergoing cardiac surgery. Others have attempted to validate this finding in the setting of cardiac surgery, with mixed success. The TRIBE-AKI Consortium published the results of the largest study investigating L-FABP in the setting of adult cardiac surgery, demonstrating that after adjustments were made for a clinical model that consisted of factors known to affect the development of AKI, L-FABP did not correlate with the development of AKI in their pediatric (n = 311) or adult (n = 1219) cohorts. The consortium demonstrated that although L-FABP levels were statistically higher in those adults with AKI than in those without AKI, the L-FABP concentration (ng/mL) measured up to 6 hours postoperatively provided an AUC of 0.61, and the performance was only marginally better with the 6- to 12-hour measurement. Similarly, in the pediatric cohort, although the fifth quintile of L-FABP concentrations at the earliest postoperative timepoint (0-6 hours) significantly associated with the development of AKI (OR, 2.9; 95% CI, 1.2 to 7.1), this effect disappeared after adjustment for the clinical model (OR, 1.8; 95% CI, 0.7 to 4.6).

Siew and colleagues reported the performance of L-FABP in 380 critically ill subjects from medical, surgical, trauma, and cardiac ICUs, in 130 of whom AKI was defined as AKIN stage 1. L-FABP levels were higher in those with AKI ( P = 0.003) and were able to discriminate incident AKI with an AUC of 0.59 (95% CI, 0.52 to 0.65). Although L-FABP was able to predict the composite endpoint of death or RRT, using multivariate regression L-FABP significantly predicted the need for acute RRT (HR, 2.36; 95% CI, 1.30 to 4.25). These findings mirror those of Doi and associates, who published a prospective single-center observational cohort study examining the performance of L-FABP in 339 mixed ICU patients. In their study, L-FABP outperformed NGAL, IL-18, NAG, and other biomarkers in the detection of AKI, defined by RIFLE Risk category. Furthermore L-FABP predicted 14-day mortality with an AUC of 0.90. This study, which followed a smaller study (n = 145) by this same group of investigators, has paved the way for L-FABP to be validated for clinical use in Japan.

In a cross-sectional study of general hospitalized patients that included 92 participants with AKI and 68 control subjects (26 healthy volunteers and 42 hospitalized: 29 patients about to undergo coronary catheterization and 13 patients in the ICU with no AKI), Ferguson colleagues demonstrated that urinary levels of L-FABP were significantly higher in subjects with AKI than in hospitalized control patients without AKI, with an AUC of 0.93 (95% CI, 0.88 to 0.97); sensitivity was 83% and specificity was 90% at a cutoff value of 47.1 ng per mg of creatinine. Nickolas and associates examined L-FABP at the time of ED arrival and found that it had only fair discriminatory power with regard to AKI. In their cohort of 1635 subjects, L-FABP provided an AUC of 0.70 (95% CI, 0.65 to 0.76); however, there was a clear and significant stepwise increase in L-FABP concentrations across the spectrum of AKI (normal < CKD < prerenal < intrinsic AKI).

Because L-FABP is also expressed by the liver, liver injury can be a potential contributor to increased urinary levels of L-FABP during AKI. However, previous studies in patients with CKD, AKI, and sepsis have shown that serum L-FABP levels do not have an influence on urinary levels and that urinary L-FABP levels are not significantly higher in patients with liver disease than in healthy subjects.

Urinary L-FABP levels have been investigated as an early diagnostic and predictive marker for contrast medium–induced nephropathy. In a study of adult patients with normal serum creatinine concentrations who underwent percutaneous coronary intervention, serum NGAL level rose at 2 and 4 hours, whereas urinary NGAL and urinary L-FABP increased significantly after 4 hours and remained elevated up to 48 hours, after cardiac catheterization. Nakamura and associates demonstrated that baseline urinary L-FABP levels were significantly higher in patients whom contrast medium–induced nephropathy developed after coronary angiography; however, the investigators did not evaluate the diagnostic performance of urinary L-FABP in predicting AKI.

Netrin-1

Netrin-1 is a 50- to 75-kDa, laminin-like protein, initially recognized as a chemotropic factor, that plays an essential role in guiding neurons and axons to their targets. Studies have now revealed diverse roles of netrin-1 beyond axonal guidance, including development of various organs, angiogenesis, adhesion, tissue morphogenesis, inflammation, and tumorigenic processes. Netrin-1 is expressed in several tissue types, including brain, lung, heart, liver, intestine, and kidney.

A study by Wang and colleagues showed a rapid induction of netrin-1 in tubular epithelial cells in response to ischemia-reperfusion injury of the kidney in animal models. In this study, netrin-1 was excreted in the urine as early as 1 hour after a kidney insult, increased more than fortyfold by 3 hours, and reached its peak levels (approximately fiftyfold) before the elevation of blood creatinine and BUN concentrations. Importantly, this rapid increase in netrin-1 expression appeared to be regulated at the translational level because netrin-1 gene transcription was actually decreased after ischemia-reperfusion injury. The researchers also tested the sensitivity and specificity of netrin-1 in animal models of toxin-induced kidney injury, using cisplatin, folic acid, and endotoxin (lipopolysaccharide). These kidney insults resulted in increases in the excretion of netrin-1 in urine, supporting a potential role as an early biomarker for hypoxic and toxic renal injuries. In a later study, through the exogenous administration of netrin-1 following a murine model of ischemia-reperfusion AKI, the same group demonstrated that netrin-1 regulates the inflammatory response in the setting of AKI via the inhibition of cyclo-oxygensase-2 (COX-2)–mediated prostaglandin E ₂ production. They demonstrated that netrin-1 regulates COX-2 expression through the regulation of nuclear factor-kappaB (NF-κB) activation.

Although most of the investigations of netrin have focused on cellular and animal models, netrin-1 has been increasingly investigated in humans. Ramesh and colleagues also demonstrated significantly higher urine levels of netrin-1 in patients with established AKI due to various causes (n = 16) than in healthy volunteers. In a later study, the same group of scientists evaluated the potential of netrin-1 to predict AKI in patients undergoing cardiopulmonary bypass. They included serial urine samples that were collected from 26 patients in whom AKI developed and 36 patients in whom it did not after cardiopulmonary bypass. By ROC analysis, the investigators demonstrated that netrin-1 could predict AKI at 2 hours, 6 hours, and 12 hours, with an AUC of 0.74 (95% CI, 0.86 to 0.89). The levels of urinary netrin-1 6 hours after cardiopulmonary bypass correlated with the severity of AKI, as well as the length of hospital stay, and remained a powerful independent predictor of AKI.

Netrin-1 seems to be a promising early biomarker for AKI, but additional studies need to be conducted in larger cohorts with AKI due to various causes to further evaluate its potential.

Neutrophil Gelatinase–Associated Lipocalin

Neutrophil gelatinase–associated lipocalin (also known as lipocalin 2 or lcn2 ) is one of the biomarkers of AKI that has been studied extensively. NGAL has many of the characteristics required for a good biomarker for AKI in comparison with serum creatinine measurement or urine output. It is a 25-kDa protein with 178 amino acids belonging to the lipocalin superfamily. Lipocalins are extracellular proteins with diverse functions involving transport of hydrophilic substances through membranes, thereby maintaining cell homeostasis. NGAL is a glycoprotein bound to matrix metalloproteinase-9 in human neutrophils. It is expressed in various tissues in the body, such as salivary glands, prostate, uterus, trachea, lung, stomach, and kidney, and its expression is markedly induced in injured epithelial cells, including those in the kidney, colon, liver, and lung.

Transcriptome profiling studies in rodent models identified NGAL as one of the most upregulated genes in the kidney very early after tubular injury. Mishra and associates demonstrated that NGAL values were significantly elevated within 2 hours after injury in mouse models of renal ischemia-reperfusion. In addition, urinary NGAL was detectable after one day of cisplatin administration, suggesting its sensitivity in other models of tubular injury.

N -Acetyl-β-D-Glucosaminidase

NAG is a lysosomal brush border enzyme that resides in the microvilli of tubular epithelial cells. Damage to these cells results in shedding of this enzyme into the urine. NAG has a high molecular weight, 130 kDa, and hence plasma NAG is not filtered by the glomeruli. Its excretion into urine correlates with tubular lysosomal activity. Increased urinary concentrations of NAG have been found in patients with AKI, chronic glomerular disease, diabetic nephropathy, exposure to nephrotoxic drugs, delayed renal allograft function, environmental exposure, contrast medium–induced nephropathy, and sepsis, and following cardiopulmonary bypass. In a prospective study involving 201 hospitalized patients with AKI, patients with higher concentrations of urinary NAG and KIM-1 were more likely to die or require dialysis. The results of this study suggest the utility of NAG in combination with KIM-1 in predicting adverse clinical outcomes in patients with AKI. In another study, urinary NAG concentrations were significantly higher in patients with contrast medium–induced nephropathy than in patients without such nephropathy within 24 hours of the administration of a contrast agent.

Similarly, in a two-center Japanese study of 77 patients undergoing cardiac surgery, NAG values were elevated in those in whom postoperative AKI developed. In this study, biomarker performance significantly improved when NAG was combined with L-FABP (an AUC improvement from 0.75 to 0.81). This same group published a single-center study investigating the performance of NAG in predicting the development of AKI (RIFLE) in a mixed, medical-surgical ICU. NAG did not perform as well in this cohort of 339 subjects, providing an AUC of 0.62 for the development of RIFLE Risk disease. In a cohort of 635 ED patients, an NAG value over 1.0 units/g provided an AUC of 0.71 (95% CI, 0.62 to 0.81) for the development of AKI during the subsequent hospital admission. However, this effect was attenuated in a multivariate analysis that included other novel and traditional biomarkers of AKI (creatinine, BUN, NGAL, etc.).

In the setting of CKD, one study of patients with type 1 diabetes and nephropathy by Vaidya and colleagues showed that lower levels of urinary KIM-1 and NAG were associated with the regression of microalbuminuria. Similarly, in a nested-case-control study from the Diabetes Control and Compliance Trial, baseline NAG concentrations were shown to predict microalbuminuria and macroalbuminuria. To date, there are little or no data on the role of NAG in CKD progression.

There are some limitations in the use of NAG as a marker of kidney injury. Inhibition of NAG enzyme activity has been reported in the presence of metal ions and at higher urea concentrations in the urine. Moreover, increased urinary levels of NAG have been reported in several nonrenal diseases, including rheumatoid arthritis and hyperthyroidism, as well as in conditions with increased lysosomal activity without cellular damage. Because of concerns about its specificity, the clinical utility of NAG as a biomarker has been limited.

Proteinuria

In a healthy person, urinary protein excretion is less than 150 mg/day and consists mainly of filtered plasma proteins (60%) and tubular Tamm-Horsfall proteins (40%). Proteinuria can result from at least three different pathophysiologic mechanisms, glomerular (increased permeability of glomerular filtration barrier to protein due to glomerulopathy, raised glomerular capillary hydrostatic pressure, or altered glomerular filtration coefficient), overflow (due to increased production of low-molecular-weight plasma proteins, e.g., immunoglobulin light chains in myeloma), and tubular processes (decreased tubular absorption of filtered proteins or increased production of tubular proteins by damaged tubules). Proteinuria mechanisms and consequences are discussed in Chapter 53 .

Proteinuria is diagnosed when total urinary protein is greater than 300 mg/24 hour. Methods for detecting and monitoring proteinuria are discussed in Chapter 26 .

Several publications highlight the diagnostic power of total protein for AKI in various drug-induced nephrotoxicities, including cisplatin and nonsteroidal antiinflammatory drugs. Low eGFR is a known risk factor for AKI, but the utility of proteinuria in combination with the eGFR to predict the risk of this disease is now being investigated. In a large cohort of nearly 1 million adult Canadians, James and colleagues demonstrated an independent association among eGFR, proteinuria, and incidence of AKI. This group reported that patients with normal eGFR levels (≥60 mL/min per 1.73 m ² ) and mild proteinuria (urine dipstick trace to 1+) have 2.5 times more risk of admission to hospital with AKI than do patients with no proteinuria. The risk was increased to 4.4-fold when patients with heavy proteinuria (urine dipstick ≥2+) were included. Adjusted rates of admission with AKI and kidney injury requiring dialysis remained high in patients with heavy dipstick proteinuria independent of the eGFR. These findings confirm previous reports suggesting that eGFR and proteinuria are potent risk factors for subsequent AKI.

Albuminuria

Albuminuria is recognized as one the most important risk factors for progression of chronic kidney diseases. Albumin is a major serum protein slightly larger than the pores of the glomerular filtration membrane, so albuminuria is best known as a biomarker of glomerular dysfunction; the appearance of albumin in large amounts in urine represents compromised integrity of the glomerular basement membrane. In smaller amounts, however, the presence of albumin in the urine may reflect tubular injury. Albuminuria is classified in the KDIGO classification system as stage A1 (urine AE [albumin excretion] < 30 mg/day or urine ACR [urine albumin to creatinine ratio] < 30 mg per g creatinine), stage A2 (previously termed microalbuminuria; urine AE 30-300 mg/day or uACR = 30-300 mg per g creatinine), and stage A3 (previously termed macroalbuminuria urine AE > 300 mg/day or urine ACR > 300 mg per g creatinine). In a number of clinical studies, albuminuria has been shown to be a sensitive biomarker of drug-induced tubular injury. It is routinely used as a marker of kidney damage for making a CKD diagnosis at eGFRs above 60 mL/min/1.73 m ² .

Guidelines of the National Kidney Foundation (NKF) and of the American Heart Association (AHA), respectively, include microalbuminuria and an increase in the urinary total protein excretion as risk factors for renal and cardiovascular diseases. Both NKF and AHA guidelines suggest measurement of urine ACR in an untimed spot urine sample. Ideally, urine ACR should be assessed in at least three different samples to decrease the intraindividual variation. Albuminuria is a continuous risk factor for ESKD and cardiovascular mortality with no lower limit, even after adjustment for eGFR and other established risk factors. Urine albumin value has been in use as a biomarker for monitoring CKD progression and to monitor potential therapeutic efficacy, although the FDA does not accept albuminuria as a surrogate marker.

Using microalbuminuria as a marker, Levin and colleagues demonstrated that N -acetylcysteine may attenuate contrast medium–induced glomerular and tubular injury. In the last several years, there has been more investigation of urine albumin excretion as a biomarker for AKI. The TRIBE-AKI Consortium measured preoperative albuminuria in 1159 adult patients, organizing the cohort into clinical risk categories on the basis of the preoperative urine ACR: 10 mg/g or less (≤1.1 mg/mmol), 11 to 29 mg/g (1.2 to 3.3 mg/mmol), 30 to 299 mg/g (3.4 to 33.8 mg/mmol), and 300 mg/g or greater (≥33.9 mg/mmol). The incidence of AKI, defined as AKIN stage 1, increased across the ACR categories, with patients whose urine ACR was greater than 300 mg/g having a relative risk of 2.36 (95% CI, 1.85 to 2.82) in comparison with to the group whose urine ACR was less than 10 mg/g. This association was slightly attenuated after adjustments for variables known to affect proteinuria and AKI (RR, 2.21; 95% CI, 1.66 to 2.73). These adult data contrast with pediatric TRIBE-AKI data (n = 294), which demonstrated no association between preoperative urine ACR and the development of postoperative AKI. The adult ACR data support use of an additional biomarker to aid in cardiac surgery AKI prediction models and supplement other data that point to the use of proteinuria/albuminuria as a biomarker of AKI both preoperatively and postoperatively.

In the postoperative setting, the TRIBE-AKI cohort showed that urine albumin concentrations (mg/L) and dipstick proteinuria values obtained within 6 hours of adult cardiac surgery correlated with the future development of AKI. Compared with the lowest quintile, the highest quintile of albuminuria and highest group of dipstick proteinuria were associated with the greatest risk of AKI (adjusted RR, 2.97; 95% CI, 1.20 to 6.91, and adjusted RR, 2.46; 95% CI, 1.16 to 4.97, respectively). However, only postoperative urine albumin concentration (mg/L) was associated with improved risk stratification when added to the clinical model (AUC increased from 0.75 to 0.81; P = 0.006). Despite its known utility in other settings, a higher early postoperative urine ACR (mg/g) was not statistically associated with AKI risk. The poor performance of urine ACR in the context of adult cardiac surgery may be explained by variations in the urine creatinine excretion within and between individuals, which could be especially prominent when renal function is not in a steady state. Urinary albumin (mg/L) in the early postoperative period was highly predictive of long-term mortality in the TRIBE-AKI adult cohort. Specifically, of patients with perioperative AKI (n = 407), those in the second tertile of albuminuria values were at increased risk of death in the 3.0-year follow-up period (adjusted HR, 2.28; 95% CI, 1.06 to 4.88). Although this effect was further magnified in the third tertile of those with AKI (adjusted HR, 2.85; 95% CI, 1.36 to 5.99); there was no increased mortality across any of the tertiles of urine albumin concentration in the 792 subjects without perioperative AKI.

In the TRIBE-AKI pediatric cohort, perioperative values of urine ACR (mg/g), and not albuminuria (mg/L), were found to be predictive of AKI. In children younger than 2 years, an absolute first postoperative urine ACR of 908 mg/g or higher (103 mg/mmol, highest tertile) predicted the development of AKIN stage 2 or 3 AKI with an adjusted RR of 3.4 (95% CI, 1.2 to 9.4) in comparison with the first tertile. In children 2 years or older, a postoperative urine ACR value of 169 mg/g or higher (19.1 mg/mmol, highest tertile), regardless of preoperative values, predicted stage 1 AKI after adjustments for clinical factors, such as age, race, sex, and preoperative eGFR, and type of cardiac surgery (adjusted RR, 2.1; 95% CI, 1.1 to 4.1). Although urine albumin concentration and urine ACR remain established and readily available laboratory tests, the diversity of results in investigations of the development of postoperative AKI indicate that further studies are needed before either may be used in clinical practice.

Urinary Cystatin C

Urinary cystatin C tracks the function of proximal tubular cells. In healthy individuals, the urinary levels of cystatin C are almost undetectable and any damage to proximal tubular cells can impede the reabsorption and enhance the urinary excretion of cystatin C. Several clinical studies sought to understand the potential of urinary cystatin C levels for prediction of kidney injury and its prognosis. Herget-Rosenthal and associates analyzed data for 85 patients in the ICU who were at high risk for development of AKI and used the RIFLE classification to define AKI. The investigators reported that serum cystatin C level signaled AKI 1 to 2 days before changes in serum creatinine level, with AUC values of 0.82 and 0.97 on day 2 and day 1, respectively, as well as demonstrating that urine cystatin C served as a marker of AKI severity, correlating with future need for RRT. Urinary cystatin C concentrations, normalized to urinary creatinine, of more than 11.3 mg/mmol were significantly associated with proteinuria. Attempts to validate urine cystatin C as a marker of ICU-associated AKI have provided mixed results. Siew and colleagues measured urine cystatin C in 380 ICU patients (mixed-surgical, medical, trauma, and cardiac) and demonstrated that there was no difference in concentrations between those with and those without AKI ( P = 0.87). More encouraging data from the EARLYARF trial showed that in a cohort of 529 subjects, urinary cystatin C may have the limited ability to detect AKI (AUC = 0.67 upon ICU arrival) with no significant difference in its ability to detect AKI in patients with GFRs above and below 60 mL/min. Additionally, in a separate post hoc analysis of the same study, urinary cystatin C values exhibited a stepwise and significant ( P < 0.001) increase in the following order: patients with no AKI, patients with prerenal AKI, patients with intrinsic AKI (which was defined as AKIN stage 1 > 48 hours).

In contrast to these mixed ICU-associated AKI data, several small studies investigating urinary cystatin C in the setting of cardiac surgery have demonstrated promise. However, these results were not validated in the TRIBE-AKI study. In unadjusted analyses of the adult cohort, several quintiles of urine cystatin C were significantly associated with the development of either mild (AKIN stage 1) or severe (doubling of creatinine or need for RRT) AKI at both the 0- to 6-hour and 6- to 12-hour postoperative time points. However, the small associations were completely attenuated after adjustments for the clinical model. Similarly, in the TRIBE pediatric cohort, no quintile remained significantly associated with AKI (mild or severe) in the adjusted analyses. Urinary cystatin C demonstrated similar results when measured in a cohort of 1635 ED patients, supplying an AUC of 0.65 (95% CI, 0.58 to 0.72) for the future development of AKI. However, utilizing a multivariate analysis that included traditional (creatinine) and more modern (urine NGAL, KIM-1, IL-18, and LFABP) biomarkers, urinary cystatin C was not a significant contributor to the prediction of the composite outcome of inpatient RRT or death. Finally, when investigated in a prospective multicenter observation cohort study of deceased-donor kidney transplants, urinary cystatin C values from the first postoperative day were modestly correlated with 3-month allograft function, whereas the AUC for predicting delayed graft function at the 6-hour postoperative timepoint was 0.69.

A number of studies have reported increased urinary cystatin C levels in patients with proteinuria, suggesting the possibility of tubular damage as a consequence of protein overload. Currently, cystatin C level has several disadvantages as a biomarker, including lack of international standardization and expense of the assay. Although serum cystatin C level has been demonstrated as a reliable biomarker of eGFR, one must remember that cystatin C synthesis is increased in smokers, patients with hyperthyroidism, those receiving glucocorticoid therapy, and those with elevations of inflammatory markers such as white blood cell count and C-reactive protein level, and the impact of these factors on urinary cystatin C in the setting of AKI has not been fully investigated. Furthermore, several different commercial assays are available to measure cystatin C. Advantages are that the commercially available immunonephelometric assay provides rapid, automated measurement of cystatin C, and results are available in minutes. In addition, preanalytic factors, such as routine clinical storage conditions, freezing and thawing cycles, and interfering substances such as bilirubin and triglycerides, do not affect cystatin C measurement.

TIMP-2 and IGFBP-7

Tissue inhibitor metalloproteinase-2 (TIMP-2) and urine insulin-like growth factor-binding protein-7 (IGFBP-7) have been shown to serve as biomarkers of AKI in the setting of critical illness. They were originally discovered as part of a three-center discovery cohort of 522 subjects. These patients had AKI stemming from sepsis, shock, major surgery, and trauma. More than 300 potential markers were evaluated, with TIMP-2 and IGFBP-7 being the two that best predicted the development of KDIGO stage 2 or 3 AKI. This finding was then validated in a prospective international multicenter observational study of 728 subjects. In this validation study, TIMP-2 and IGFBP-7 remained the top two performing biomarkers for the prediction of RIFLE Injury or Failure stage within the first 12 to 36 hours of study enrollment, providing AUCs of 0.77 and 0.75, respectively. When these two biomarker values were multiplied together, they demonstrated an improved ability to detect this same endpoint (AUC = 0.80). However, we are cautious about the interpretation of these data because the study did not supply information about the combination of TIMP-2 and IGFBP-7 with other biomarkers of AKI (e.g., NGAL, KIM-1, L-FABP, IL-18). This is the first and only publication to demonstrate that either of these markers plays a role in diagnosing AKI and will require validation in follow-up studies. These biomarkers are unique in that they play a role in cell cycle arrest. Both IGFBP-7 (through p523 and p21) and TIMP-2 (through p27) block the effect of cyclin-dependent protein kinase complexes, resulting in short periods of G ₁ cell cycle arrest. A commercial assay for these biomarkers is currently available in Europe, but the TIMP-2 and IGFBP-7 test has not been approved for clinical use in the United States.