Past, present, and future biomarkers of kidney function and injury: The relationship with antibiotics

Abstract

Routinely used kidney biomarkers of injury and function such as serum creatinine and urine albumin to creatinine ratio, are neither sensitive nor specific. Future biomarkers are being developed for clinical use and have already been included in guidance from groups such as the U.S. Food and Drug Administration and the Predictive Safety Testing Consortium. These biomarkers have important implications for early identification of kidney injury and more accurate measurement of kidney function. Many antibiotics are either eliminated by the kidney or can cause clinically significant nephrotoxicity. As a result, clinicians should be familiar with new biomarkers of kidney function and injury, their place in clinical practice, and applications for antibiotic dosing.

1. Background

Acute kidney injury (AKI) is a frequent complication among hospitalized patients, which is associated with significant morbidity and mortality [1–3]. AKI is characterized by a sudden decrease in the excretory function of the kidney, which results in the accumulation of nitrogenous waste products, altered clearance of various xenobiotics, and electrolyte dysregulation [4]. Although the incidence of AKI varies widely across different populations and settings of care, nephrotoxic medications are one of the few modifiable risk factors for AKI [5]. Resulting kidney impairment is generally reversible upon discontinuation of the offending medication; however, medication-induced AKI causes increased morbidity, hospitalizations, and healthcare-related costs [6]. Approximately 20% of hospital-associated AKIs are due to medications, with rates up to 60% observed in older adults [3,7–10]. Anti-infectives represent one of the most common causes of AKI, and are attributed to nearly 30% of cases of medication-induced nephrotoxicity [11]. Commonly used antibiotics and antivirals such as aminoglycosides, β-lactams, polymyxins, trimethoprim–sulfamethoxazole, vancomycin, acyclovir, and foscarnet have all been associated with significant rates of medication-induced nephrotoxicity.

Serum creatinine (SCr) remains the primary clinical biomarker for monitoring of kidney function. In particular, SCr plays a critical role in detection of AKI and dose adjustments for renally cleared medications; however, SCr is a non-specific biomarker which is insufficiently sensitive for the early detection of acute decreases in glomerular filtration rate (GFR), and does not adequately identify specific types of kidney injury [12]. By using non-specific biomarkers of kidney injury such as SCr, clinicians are limited to identification of ‘established AKI’ and ‘no AKI’ (Fig. 1). This is problematic, since patients may transition between these 2 criteria via the intermediate stages of ‘subclinical AKI,’ in which kidney damage occurs prior to realized GFR changes, and ‘pseudo nephrotoxicity,’ in which SCr becomes elevated in the absence of true kidney injury [13]. Furthermore, there are currently no clinically qualified biomarkers to identify the presence of drug-induced kidney injury in humans. Because medications are one of the few modifiable risk factors for AKI, more sensitive and specific biomarkers of kidney injury and kidney function may be helpful for early identification of kidney injury [14,15].

Fig. 1.

Continuum of glomerular filtration rate (GFR) and kidney injury biomarkers across various stages of acute kidney injury (AKI). Adapted from Cruz and Mehta [13].

In the past decade, significant emphasis has been placed on translational research related to specific biomarkers to better characterize GFR and kidney injury [16–18]. First, in 2008, urinary kidney injury biomarkers including kidney injury molecule–1 (KIM-1), cystatin C, and clusterin were qualified by the United States Food and Drug Administration (FDA), the European Medicines Agency, and the Pharmaceuticals and Medical Devices Agency (PMDA) for the detection of drug-induced nephrotoxicity in rats. Regulatory agencies recommended these urinary biomarkers to be used in conjunction with traditional measures of nephrotoxicity such as blood urea nitrogen (BUN), SCr, and histopathological changes, to aid in the detection of acute kidney tubular changes in Good Laboratory Practice (GLP) rat studies used to support clinical trials [19–21]. In 2018, the FDA awarded the first qualification of a clinical safety biomarker to the Foundation for the National Institutes of Health Biomarkers Consortium and the Critical Path Institute Predictive Safety Testing Consortium [22]. This qualification was for a composite measure of 6 urinary biomarkers including clusterin, cystatin-C (cysC), KIM-1, N-acetyl-β-D-glucosaminidase (NAG), neutrophil gelatinase–associated lipocalin (NGAL), and osteopontin (OPN) [23]. Currently, the FDA recommends this composite panel of urinary biomarkers to be used in conjunction with traditional measures of kidney injury in phase 1 trials in healthy volunteers. Since the qualification of these novel urinary biomarkers by regulatory agencies, use in drug development and research programs has increased [24,25]. However, these urinary biomarkers are not yet qualified for clinical use, and more data are needed regarding their performance in different disease states, patient populations, and responses to differing types of AKI.

2. Main text

Over the last several decades, evolving definitions of acute kidney injury (AKI) have been proposed and utilized [26–28]. Collective efforts culminated in the 2012 Kidney Diseases Improving Global Outcomes (KDIGO) Working Group criteria, which established a unified definition and staging criteria for AKI in clinical practice and research [27]. Despite the creation of these unified criteria, the current KDIGO guidelines are still significantly limited by the use of serum creatinine (Scr) and urinary output as the primary criteria for AKI identification and stratification.

2.1. Types of medication-induced kidney injury

Drug-induced nephrotoxicity occurs by multiple pathogenic mechanisms including tubular cell toxicity, inflammation, crystal nephropathy, rhabdomyolysis, thrombotic microangiopathy, and altered intraglomerular hemodynamics [29]. Tubular cell toxicity occurs when renal tubular cells (most commonly proximal tubular cells) are exposed to high levels of circulating nephrotoxins [30–32]. Mitochondrial function is impaired, which interferes with tubular transport, causing oxidative stress and formation of free radicals [33]. Importantly, drug-induced tubular cell toxicity occurs via several different mechanisms: (1) proximal tubular injury due to accumulation of drugs or their metabolites (e.g., aminoglycosides, vancomycin, amphotericin B, cidofovir, cisplatin) [33–35]; (2) tubular obstruction resulting from crystals or casts formed by drugs or their metabolites (e.g., vancomycin, acyclovir, foscarnet, ciprofloxacin) [34,36]; and (3) Fanconi syndrome, or a generalized transport defect in the proximal tubules, which can be precipitated by medications such as aminoglycosides, tetracyclines, and antivirals such as tenofovir, cidofovir, and didanosine [37,38] (Fig. 2).

Fig. 2.

Information in blue-shaded boxes represents specific biomarkers of kidney function and injury at their respective locations within the kidney. Selected nephrotoxins and associated sites of nephrotoxicity are highlighted in orange-shaded boxes.

Medications can also cause inflammatory changes within the glomerulus, renal tubular cells, and renal interstitium, which leads to fibrosis [31]. Anti-infectives are most often associated with acute interstitial nephritis (AIN), an idiosyncratic, allergic response which develops in a non–dose-dependent fashion. The underlying mechanism is thought to be caused by a drug or metabolite which binds to the tubular cell membrane or renal interstitium, thus acting as an antigen and causing an immune response that results in AIN [32]. Examples of anti-infective agents that can cause AIN include β-lactam antibiotics, fluoroquinolones, sulfonamides, acyclovir, and rifampin [32,33].

Another mechanism of drug-induced nephrotoxicity is crystal nephropathy. This results from use of medications with crystalforming capabilities, which can supersaturate in the urine with volume depletion, dehydration, and excessive drug dosing. Precipitation of crystals within the tubular lumen leads to obstruction of urinary flow, which elicits an interstitial reaction and impairs kidney function [31]. Anti-infective medications that can cause crystal nephropathy include aminopenicillin antibiotics, sulfonamides, atazanavir, darunavir, acyclovir, ganciclovir, and foscarnet [35,39].

Rhabdomyolysis, or skeletal muscle injury leading to lysis of myocytes, is a less common cause of drug-induced nephrotoxicity [40]. While anti-infectives are a rare cause of this adverse effect, interactions between select agents such as daptomycin and macrolides with HMG-CoA reductase inhibitors (statins) can lead to elevated risk of rhabdomyolysis. Daptomycin is a lipopeptide antibiotic for Gram-positive bacteria which is known to cause skeletal muscle breakdown and corresponding creatinine phosphokinase (CPK) elevations. The reported incidence of rhabdomyolysis in phase III clinical trials was 2.1%, and few case reports have reported this toxicity since the drug was approved by the FDA in 2003 [41–43]. Although case reports are limited, common patient factors in cases of daptomycin-associated rhabdomyolysis include elevated trough exposures (C_min ≥ 20 μg/mL), dosing intervals <24 h, kidney dysfunction, and concomitant administration of HMG-CoA reductase inhibitors [40,44,45]. Macrolides are an uncommon cause of rhabdomyolysis; however, case reports of rhabdomyolysis resulting from interactions between select agents such as clarithromycin and erythromycin with statins such as simvastatin and lovastatin have been reported [46,47].

Current management for drug-induced nephrotoxicity involves recognition of the adverse event, removal or adjustment of the offending agent, steroid treatment, and renal replacement therapy as necessary [31,48,49]. Since many instances of drug-induced nephrotoxicity are reversible with prompt management, efforts have shifted towards identifying early and specific biomarkers for kidney damage.

2.2. Markers of kidney function

According to the most recent FDA guidance document on determination of kidney function in pharmacokinetic (PK) studies, exogenous markers of GFR such as inulin, iothalamate, and iohexol are more accurate measures of true GFR than SCr-based estimation equations [50]. However, the agency remains agnostic on the equations that are ultimately applied for dosing guidance in the drug label. Given the current landscape of clinical practice where SCr-based equations remain the standard, SCr-based estimated glomerular filtration rate (eGFR) remains the most used tool to estimate renal elimination of drugs in PK studies. Alternative biomarkers such as cystatin C have been proposed, but these will require integration in pre-clinical and clinical studies for future application in practice. Kidney function biomarkers with improved accuracy over SCr-based methods are needed, since proper estimation of GFR is a key element of diagnosis, drug dosing, and other critical aspects of patient care (Table 2). It should be noted that many of these specialized biomarkers, such as cystatin C and iohexol, are not intended for use across all patients but rather in specific subsets of patients more susceptible to kidney injury who require closer monitoring of kidney function, or when traditional methods may be inaccurate.

Table 2.

Summary table of kidney function and kidney injury biomarkers.

Biomarker	Location	Function	Utility	Limitations
Kidney function biomarkers
Serum creatinine (SCr)	Freely filtered by the glomerulus and actively secreted by transporters in the proximal tubules	Used for creatinine clearance calculations as a surrogate for GFR	Most commonly utilized clinical assessment for GFR due to low cost and decades of clinician familiarity	Measurements are highly susceptible to variations in SCr due to individual patient factors
Inulin	Freely filtered by the glomerulus and neither secreted nor reabsorbed in kidney tubules	Used for inulin clearance calculations as a surrogate for GFR	Gold standard marker for kidney function, but use is primarily limited to research settings	Measurements are labor and resource intensive, limiting widespread clinical implementation
Iohexol	Freely filtered by the glomerulus and neither secreted nor reabsorbed in kidney tubules	Used for iohexol clearance calculations as a surrogate for GFR	Retains many characteristics of a gold standard functional biomarker (free filtered, non-secreted, non-metabolized) with improved ease of use in clinical and research settings	Clinical use has largely been limited to specific regions such as Europe Lack of a standardized equation for iohexol–GFR calculation limits broader clinical implementation
Fluorescein isothiocyanate–conjugated sinistrin (FITC-S)	Freely filtered by the glomerulus and neither secreted nor reabsorbed in kidney tubules	Plasma clearance can be measured transdermally as a surrogate for GFR	Retains many characteristics of a gold standard functional biomarker (free filtered, non-secreted, non-metabolized) with improved ease of use in clinical and research settings	Use and experience have primarily been limited to research settings, but clinical trials are underway
Serum cystatin C (cysC)	Freely filtered by the glomerulus, then reabsorbed and catabolized by the renal tubular epithelium	Used for cysC clearance calculations as a surrogate for GFR	Has improved characteristics (freely filtered) compared to the current clinical standard of SCr; can provide more accurate GFR estimation when SCr is affected by certain individual patient-level factors	Measurements may be affected by individual patient factors (obesity, malignancy, corticosteroid use)
Kidney injury biomarkers
Serum creatinine (SCr)	Freely filtered by the glomerulus and actively secreted by transporters in the proximal tubules	Changes in SCr are utilized for AKI recognition and grading	Most commonly utilized clinical assessment for AKI due to low cost and decades of clinician familiarity	Detectable rise in SCr is susceptible to significant delays (48–72 h) after an initial AKI event
Neutrophil gelatinase–associated lipocalin (NGAL)	Normally expressed throughout the body at low levels; freely filtered by the glomerulus and reabsorbed in the proximal tubules	Increased concentrations in response to ischaemic and toxic tubular injury	Qualified for use as a composite urinary biomarker panel for detection of kidney injury as an adjunct to traditional SCr-based methods phase 1 trials. Upcoming commercial availability for identification of AKI in pediatric patients (≥3 mo to <22 y) in 2024 as ProNephro AKI™	Measurements may be affected by individual patient factors such as inflammation due to disease states; more clinical experience required across adult patient populations and different disease states
Kidney injury molecule–1 (KIM-1)	Normally expressed at low levels (<1 mg/mL) by kidney cells	Increased concentrations in response to proximal tubular injury	Qualified for use in preclinical rat studies, and as a composite urinary biomarker panel for detection of kidney injury as an adjunct to traditional SCr-based methods in phase 1 trials; significant preclinical experience in identification of vancomycin-induced nephrotoxicity	More clinical experience required across adult patient populations and different disease states
Clusterin	Normally expressed throughout the body, but at higher concentrations in undifferentiated cells in the kidney tubules	Increased concentrations in response to drug-induced nephrotoxicity and ischaemic kidney injury	Qualified for use in preclinical GLP rat studies, and as a composite urinary biomarker panel for detection of kidney injury as an adjunct to traditional SCr-based methods phase 1 trials	More clinical experience required across adult patient populations and different disease states
Osteopontin	Expressed by different tissues throughout the body, and multiple areas within the kidney such as the glomerulus and tubules	Increased concentrations in response to drug-induced nephrotoxicity and tubular injury	Qualified for use as a composite urinary biomarker panel for detection of kidney injury as an adjunct to traditional SCr-based methods phase 1 trials	Measurements may be affected by individual patient factors such as cancer and chronic kidney disease
Tissue inhibitor of metalloproteinase 2 (TIMP-2) and insulin-like growth factor binding protein 7 (IGFBP7)	Normally expressed by the kidneys and other body tissues at low levels	Expressed in renal tubular cells during acute cellular stress or injury	Available commercially as NephroCheck®, which is approved as an aid for AKI risk assessment in adult ICU patients; significant clinical experience when used for AKI detection among cardiac surgery, sepsis, and/or hemodynamic instability	Limited clinical experience when used specifically for drug-induced nephrotoxicity

Abbreviations: AKI, acute kidney injury; cysC, serum cystatin C; GFR, glomerular filtration rate; GLP, Good Laboratory Practice; ICU, intensive care unit; SCr, serum creatinine.

2.3. Creatinine

Creatinine is a 113-Da molecule that is produced as a normal, endogenous product of muscle catabolism. This occurs as a continuous process whereby creatine in the muscle undergoes nonenzymatic conversion to creatinine and is then excreted by the kidneys [35,51,52]. In normal patients, creatinine is produced at a relatively stable rate, as a function of their lean muscle mass. As a result, SCr remains widely used in clinical practice for both estimation of GFR and identification of AKI [27,53–55].

Use of SCr to estimate GFR is notably influenced by non-renal factors at the patient level, such as age, sex, diet, muscle mass, pregnancy, concurrent medications, and coexisting disease states [56,57]. Equations attempt to account for some of these factors, but inaccuracies persist in GFR estimation and thus identification and staging of AKI [58,59]. Another significant limitation to the clinical use of SCr is the delay in detectable rises after an AKI event. Because the half-life of SCr increases inversely to declining GFR, there is typically a delay of 48–72 h after an acute change in kidney function, before a detectable rise and peak in SCr occur [60–63]. Clinically, this can result in significant delays between the time of initial kidney injury and subsequent detection and management of AKI.

Creatinine undergoes free filtration at a constant rate in the glomerulus but is also secreted by various active transporters within the kidney tubules. These include organic cation transporter 2 (basolateral uptake of creatinine in proximal tubular cells) and multidrug and toxin extrusion efflux proteins (apical efflux of creatinine in proximal tubular cells) [56]. Use of certain medications that inhibit the tubular secretion of creatinine can also lead to rapid increases in SCr of approximately 15%–30%, which are ultimately reversible, and independent of any changes in true GFR [64]. Examples of these medications include trimethoprim, cimetidine, cobicistat, and dolutegravir, among others [65] (Table 1).

Table 1.

Representative data for common drugs which compete for creatinine secretion.

Drug	Increase is steady-state creatinine	Population	References
Trimethoprim	0.22 mg/dL	2 Female patients and 2 healthy male patients	[66]
Trimethoprim	0.20 ± 0.13 mg/dL (22.2% increase) and 0.28 ± 0.18 mg/dL (31.3% increase)	10 Healthy female and 10 healthy male patients receiving moderate (766.7 ± 143.5 mg) and high (1600.0 ± 313.8 mg) doses of trimethorprim	[67]
Cimetidine	0.68 (26%), 0.29 (23%), and 0.18 mg/dL (19%) for those with estimated GFR of <40, 40–80, and >80 mL/min/1.73 m2	Patients with varying kidney function	[68]
Cimetidine	Scr change of 0.14 mg/dL	18 Patients with active duodenal ulceration shown by fibreoptic endoscopy	[69]
Cimetidine	SCr change from 0.96 mg/dL to 1.12 mg/dL (0.16 mg/dL difference) in those receiving 1.6 g of cimetidine	13 Patients with endoscopically proven duodenal ulcer	[70]
Cobiscistat	0.14 mg/dL	348 treatment- naive HIV patients	[71]
Dolutegravir	0.12–0.15 mg/dL	414 HIV patients	[72]

Abbreviations: GFR, glomerular filtration rate; SCr, serum creatinine.

2.4. Inulin (GFR)

Inulin is an inert, freely filtered, exogenous polysaccharide that is neither secreted nor reabsorbed in the tubules, and is neither synthesized nor metabolized by the kidney [73,74]. These properties have led to inulin being classically referred to as a gold standard for measuring GFR [75–77]. Despite its status as a gold standard biomarker for GFR measurement, inulin is not commonly used in clinical practice. Measurement of urinary inulin clearance is labor intensive, requiring a continuous intravenous infusion of inulin and urinary catheterization, followed by timed urinary and plasma collections [78].

In order to simplify the measurement of inulin clearance, alternative methods such as a single-bolus injection of inulin have been proposed [79,80]. A single-bolus technique is a faster and less labor-intensive alternative method. However, this still requires multiple blood samples (or Bayesian priors) and has been noted in the past to be less accurate than continuous infusion, with bias towards overestimating GFR. Swinkel et al attempted to improve the single-bolus technique by analyzing serum inulin time–decay curves in 59 children with impaired kidney function (GFR <60 mL/min/1.73 m²). In this study, a single dose of intravenous inulin was administered, and 11 timed blood samples were then collected between 0 and 240 min after the injection. A 2-compartment model best described the GFR in children using the serum inulin disappearance curve between 0 and 240 min. Notably, it was found that reducing the sampling frequency from 11 to 6 plasma timepoints did not significantly change the estimated GFR (mean [SEM]: 46.1 [3.0] vs. 46.7 [3.1] mL/min/1.73 m²) [79]. Further investigations of the single-bolus technique led to more refined protocols with decreased plasma sampling. Van Rossum et al were able to achieve comparable GFR estimates between single-bolus versus continuous infusion, using only 4 plasma samples. The average difference in estimated GFR between the 2 methods was 9.7 mL/min/1.73 m^2, with a greater difference observed at higher GFRs (> 100 mL/min/1.73 m²) [80].

Despite efforts to improve its clinical utility, injectable inulin is currently unavailable in the United States, and many clinical laboratories do not maintain measurement capabilities [81]. Laboratory quantification of inulin remains complex and labor-intensive; colorimetric methods first developed in the 1940s used reagents such as diphenylamine, resorcinol, and anthrone in multistep processes involving reagent preparation, reaction with inulin, and preparation of urine/plasma samples [12,82,83]. These original methods were subject to inaccuracies and inter-laboratory variation, in addition to requiring significant patient sample volume (2–4 mL of plasma or urine). Later methods simplified this process by use of enzymatic determination of inulin [84,85]. These enzymatic methods reduced variability associated with reagent preparation, and required much lower urine/plasma sample volumes (0.1–0.5 mL). Despite these improvements, determination of inulin by enzymatic methods remains more complex than possible for most standard clinical applications. Consequently, use of inulin remains limited primarily to research settings. Contemporary efforts have focused on identifying alternative molecules that share the characteristics of this gold standard biomarker for kidney function, but with improved ease of use in clinical settings.

2.5. Iohexol (GFR)

Iohexol is an intravenous, low-cost, non-ionic contrast agent which is widely used for medical imaging. The molecule has a number of characteristics that are desired for an ideal marker of GFR; it is freely filtered and eliminated by the kidneys, non-secreted, non-reabsorbed, and non-metabolized, with low plasma protein binding [86,87]. Iohexol clearance is well described by a 2-compartment PK model. Following intravenous bolus injection of iohexol, plasma concentrations decrease at a constant rate that is defined by 2 different exponential curves [87,88]. The first, rapid elimination curve corresponds to the alpha distribution phase in which distribution and clearance of iohexol occur simultaneously. The second, slower elimination curve corresponds to the beta elimination phase in which distribution has equilibrated and plasma iohexol is cleared by the kidneys [89]. For simplification of analysis and iohexol clearance calculations, a 1-compartment model of plasma clearance can be used by sampling during the terminal elimination phase. In this case, as few as 1 plasma iohexol level can still provide equivalent GFR estimates, compared to when multiple plasma levels are used [90,91]. It should be noted, however, that when this method is used, timing of plasma samples is important, in order to allow completion of the alpha distribution phase. When kidney function is normal, plasma samples should be obtained at least 2 hours post iohexol injection; when kidney function is poor, plasma samples should be obtained at least 4 h post iohexol injection [92,93].

Clinical experience, particularly in Europe, has also been accumulating regarding the safety and efficacy of iohexol [92,94,95]. The safety of iohexol has been extensively studied in several large cohorts. A retrospective summary of ~8000 GFR determinations by iohexol at a Swedish hospital over a 10-year period found that no serious adverse events, including anaphylaxis, occurred over the study period [96]. Another review of 25 years of clinical experience with iohexol-determined GFR at an Italian center showed a similar safety profile. Among nearly 3000 patients with more than 15 000 GFR determinations by iohexol, only 1 patient experienced a moderate-grade adverse event (flushing, urticaria, and itching) [97]. Notably, there were no adverse renal events noted, likely in part explained by the low doses of iohexol that are typically administered for GFR monitoring (5–10 mL for GFR monitoring vs. 80–180 mL for imaging applications [300 mg iohexol/mL]) [94,96–98].

Another key advantage of iohexol is that plasma levels can be easily quantified using liquid chromatography with tandem mass spectrometry (LCMS-MS) [99,100]. LCMS-MS is widely available in reference laboratories and allows for quantification of iohexol using minimal amounts of plasma (50–200 μL) [92]. Furthermore, iohexol remains stable at room temperature, −20°C, and −80°C [101]. This is important for its utility in clinical settings, as samples can be easily stored and transported prior to analysis. To calculate clearance, iohexol can be treated as any xenobiotic modeled according to standard mass transit pharmacokinetic models. Identification of Bayesian priors is possible, and Bayesian posteriors can reduce the number of samples needed to precisely identify patient GFR at the exact time that the assay was performed.

Iohexol is a cost-effective and utilitarian way to monitor GFR compared to traditional gold standard biomarkers such as inulin, and several studies have compared the performance of these biomarkers. In a study among 150 pediatric patients (median age: 4.6 years) mostly with normal kidney function (inulin-GFR: 75–120 mL/min/1.73m²), Lindblad and Berg compared GFRs determined by either iohexol or inulin clearance. It was found that iohexol-estimated GFR correlated well with inulin-estimated GFR (r = 0.834). Correlation between iohexol-estimated GFR and calculated creatinine clearance (CrCL) was worse but reasonable (r = 0.672). Of note, the authors found that calculated iohexol clearances were comparable, whether calculated with 3- or 1-plasma sample(s). A separate study among 20 healthy adults with normal kidney function (median inulin-GFR: 117 [IQR: 106–129] mL/min/1.73m²) compared inulin-determined GFR (gold standard) with iohexol and 4 different creatinine clearance equations. Plasma clearances for inulin and iohexol were determined either by dividing the administered dose of the respective biomarker by the area under the plasma concentration–time curve (using 16 plasma samples), the slope clearance method according to Brochner–Mortensen (using 4 or 5 plasma samples), or the Jacobsson formula, based on the administered dose of the biomarker, time between dose and blood sampling, plasma concentration of the respective biomarker, and volume of distribution [91,102]. CrCL was calculated by the 4 different equations was consistently found to be lower, compared to the gold standard inulin clearance (median difference: 23 mL/min/1.73m²). No differences between the clearances of inulin and iohexol were identified when using 16-, 5-, or a single-plasma sample(s), following administration of iohexol as an intravenous bolus.

Iohexol is an attractive option as a relatively low-cost clinical alternative for monitoring GFR. In particular, iohexol may provide more accurate GFR readings over traditional biomarkers in clinical settings where SCr may be altered due to individual patient factors. Examples of patient populations who may benefit include critically ill, elderly, and obese patients. Potential barriers to widespread clinical adoption include lack of a standardized clinical equation to calculate GFR, unfamiliarity amongst clinicians, and increased cost compared to traditional measures.

2.6. Fluorescein isothiocyanate–conjugated sinistrin (GFR)

Fluorescein isothiocyanate–conjugated sinistrin (FITC-S) is an inulin analog that is a gold standard intravenous, exogenous marker for measurement of GFR. FITC-S was introduced after fluorescein isothiocyanate–labeled inulin (FITC-I) was initially developed as a GFR biomarker in rats [103]. Although FITC-I is also considered a gold standard biomarker of kidney function, its utility in clinical practice is limited by poor water solubility and the need to collect extensive blood and urine [103–105]. In contrast to FITC-I, FITC-S clearance is monitored with a noninvasive transcutaneous monitoring device which excites FITC-S at 480 nm and detects emitted light through the skin at 520 nm [106]. The primary advantage of transcutaneous measurement of FITC-S clearance is that it offers a precise, real-time assessment of GFR without the need for blood or urine sampling [104,106].

In preclinical animal models, FITC-S has been shown to be an effective, real-time marker of GFR changes associated with vancomycin-induced kidney injury [107,108]. In a pilot study for validation of the transdermal FITC-S technique for GFR measurement, male Sprague–Dawley rats received either intravenous vancomycin (VAN) or saline, immediately followed by intravenous FITC-S and transdermal fluorescence monitoring, for 3 days. Mean GFR among rats that received saline was found to remain stable at 1.2 ± 0.04 mL/min/100 g body weight over the course of the study. Rats that received an allometrically scaled VAN dose of 150 mg/kg/day experienced an approximately 50% decline in GFR immediately after receipt of the first VAN dose on day 1 [107]. A subsequent study in male Sprague–Dawley rats utilized intravenous FITC-S with transdermal monitoring to investigate real-time GFR changes associated with the combination of vancomycin plus piperacillin–tazobactam. Similar to the validation study, rats received allometrically scaled doses of intravenous VAN, with or without intraperitoneal piperacillin–tazobactam, immediately followed by intravenous FITC-S and transdermal fluorescence monitoring, for 4 days. Changes in kidney injury via biomarkers were immediately detected following administration of study drug treatment, with significant changes in GFR identified on day 4 among rats that received VAN only [108].

These preclinical studies provide preliminary data that support the use of intravenous FITC-S and transdermal monitoring for estimation of real-time GFR in patients receiving nephrotoxic antibiotics. The ability to obtain immediate, continuous results using FITC-S provides a valuable use for this in patients who are at risk for AKI, and, when utilized, can be used to help direct patient care to minimize further damage. Translation is limited by the fact that special monitoring devices are needed for each subject studied, and monitoring is a dedicated activity.

2.7. Serum cystatin C (GFR)

Cystatin C (CysC) is a 13-kDa cysteine protease inhibitor that is secreted from all nucleated cells [109]. CysC is freely filtered by the glomerulus, then reabsorbed and catabolized by the renal tubular epithelium [110]. Compared to SCr, it is produced at a constant rate across populations and is less affected by age, sex, race, muscle mass, and diet [111]. CysC was first discovered in the 1920s, and has been investigated as a marker for GFR over the decades [112]. In a 2012 landmark trial by Inker et al, it was shown that serum cysC in combination with SCr resulted in more accurate estimations of GFR, compared to each biomarker alone [113]. Importantly, this study utilized a newly certified reference material for cysC, established by the International Federation of Clinical Chemistry Working Group for the Standardization of cysC and the Institute for Reference Materials and Measurements, in developing the new eGFR equations [114]. This was a notable standardization, since different reference materials for cysC from various manufacturers were used in studies prior to this, making extrapolation of results difficult [115]. Around this time, Peralta et al used a population-based cohort of U.S. adults in order to reclassify chronic kidney disease (CKD) according to cysC-based eGFR [116]. This study compared the association of reduced eGFR (<60 mL/min/1.73m²) as defined by SCr and cysC, with the longitudinal risk for end-stage renal disease (ESRD) or all-cause mortality. Compared to classification of CKD by SCr and albumin-to-creatinine ratio, addition of cysC criteria (either <60 or ≥60 mL/min/1.73m²) were found to improve the predictive accuracy for ESRD and all-cause mortality.

Based on these findings, the 2012 Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) eGFR_cystatinC and eGFR_{creatinine-cystatinC} equations were subsequently developed and incorporated into the KDIGO 2012 guidelines [2]. Of note, the 2012 CKD-EPI eGFR_{creatinine-cystatinC} equation resulted in the best GFR estimations, while still including age, sex, and race as factors [113,117,118]. Most recently, in 2021, CKD-EPI developed new eGFR equations that use SCr and cysC but omit race as a factor. The primary goals of the updated raceless eGFR equations were to minimize inaccuracies of GFR predictions for non-Black and Black patients, in addition to reducing differences in eGFR between race groups [95]. Prior eGFR equations that include race as a factor may overestimate GFR in Black patients, which can subsequently result in underdiagnosis of conditions such as CKD, exposure to inappropriate medications and interventions contraindicated at lower GFR, and delayed access to kidney transplantation evaluation.

Despite these apparent advantages of cysC-guided GFR estimation, several limitations should be considered. Notably, serum CysC can still be affected by other, non-renal factors such as asthma, obesity, hyperthyroidism, malignancy, HIV infection, and corticosteroid use [110]. Successful implementation of cysC-based GFR monitoring will also require drug-specific dosing recommendations. Current evidence for renal dose adjustments of medications are based on CrCL cutoffs, which do not always correlate well with cysC-measured GFR. In 1 study of 308 hospitalized adults who had both SCr and serum cysC measured, significant discordance in resulting dose adjustments were seen when cysC-based GFR equations were used in place of SCr-based GFR equations [119]. Consequently, widespread clinical implementation of cysC-based estimates of GFR, with and without creatinine, will require careful validation of the biomarker’s performance in specific patient populations, along with establishment of cysC-based renal dose adjustment cutoffs based on observed PK.

In conjunction with the need for development of new cysC-based drug dosing recommendations, more studies exploring the predictive accuracy of cysC on specific drug clearance are needed. One systematic review of 28 studies published between 2004 and 2017 identified a total of 3500 patients who were treated with 16 various renally eliminated medications [117]. The majority of the studies evaluated antimicrobials such as vancomycin, aminoglycosides, and β-lactams; overall, cysC-based estimates of GFR were an equivalent or better predictor of drug clearance, compared to SCr-based methods. Miano et al. utilized cysC in a recent clinical study to investigate potential pseudonephrotoxicity associated with the combination of piperacillin–tazobactam and vancomycin (TZP) [120]. This antibiotic combination was found to be associated with a higher incidence of creatinine-defined AKI, but was not associated with changes in cysC measured GFR; additionally, TZP was not associated with higher rates of dialysis or 30-day mortality. In this clinical example, cysC was a better measure of true GFR compared to SCr, due to the potential for inhibition of tubular re-absorption of creatinine by TZP [121,122].

2.8. Biomarkers of kidney injury or damage

Most nephrotoxic medications cause damage at primary sites within the glomerulus or proximal tubule. Consequently, use of more sensitive kidney injury biomarkers which are expressed in response to specific types or regions of kidney injury (Fig. 2) will allow for identification of different types of kidney injury [123] (Table 2). Although some biomarkers such as NGAL and TIMP-2/IGFBP7 have validated clinical cutoffs, these are typically specific to age ranges and patient populations. Because different nephrotoxicants and pathologies can result in differing types of kidney injury at various sites, these urinary injury biomarkers may ultimately be more informative when used as a panel. Changes in expression patterns of injury biomarkers could then be monitored over time to stage AKI risk, and/or identify AKI.

2.9. Neutrophil gelatinase–associated lipocalin (injury)

Neutrophil gelatinase–associated lipocalin (NGAL) is a 25-kDa, lipocalin-type protein which is expressed at consistent, low levels by a variety of cells in the uterus, prostate, lungs, colon, and kidney [124]. At baseline physiologic conditions, circulating plasma NGAL is filtered by the glomerulus and reabsorbed in the proximal tubules. This results in very low levels of urinary NGAL at baseline (<5 ng/mL) [125]. The physiologic role of NGAL is 2-fold: (1) it exerts a bacteriostatic effect by binding and sequestering iron–siderophore complexes, and (2) it provides antiapoptotic and enhanced proliferation effects on kidney tubular cells after ischemic kidney injury [126,127]. Although NGAL is expressed in multiple organs, it is markedly upregulated soon after acute kidney injury in a variety of animal and human studies [128,129]. During acute proximal tubule injury, NGAL synthesis from the kidneys and other organs may be increased, while reabsorption in the tubules simultaneously decreases, leading to detectable rise in urinary NGAL levels. More recent data have localized NGAL production following ischaemic or toxic kidney injury to the thick ascending limb of the loop of Henle and collecting duct cells, thus indicating its potential role as an early marker of structural kidney tubular damage caused by nephrotoxic medications [129],

Approximately 20 years ago, the first animal studies investigating NGAL expression in response to kidney injury were conducted in rats and mice. Several mechanisms of kidney injury were investigated, including ischemia/reperfusion injury, cisplatin-induced, aminoglycoside-induced, and mercuric chloride–induced kidney injury [130–133]. These early studies showed that NGAL was markedly increased and detectable within 1 h after onset of AKIm earlier than other urinary biomarkers of tubular injury such as N-acetyl-β-D-glucosaminidase and β−2 microglobulin.

Most clinical data for NGAL have been derived from patients undergoing cardiac surgery and those who were critically ill. One of the first studies of urinary NGAL as an early, noninvasive biomarker for acute kidney injury was conducted in pediatric patients undergoing cardiac surgery. Results were promising, since large elevations in both urinary and serum NGAL were detectable up to 48 hours prior to AKI detection using SCr. Urinary NGAL concentrations measured 2 h after cardiopulmonary bypass was also found to be a near-perfect predictor of AKI in this cohort (area under the receiver operating characteristic curve: 0.998) [134]. Further investigation of the predictive performance of NGAL has yielded more heterogeneous results, particularly among adult critically ill patients. Studies among adult patients in the intensive care unit (ICU) which investigated the predictive ability of urinary NGAL for moderately severe AKI (according to RIFLE or AKIN criteria) have yielded area under the receiver operating characteristic curve values ranging from 0.54 (close to no useful information) to 0.99 (excellent predictive performance) [135–138]. Several reasons may explain these findings. Critically ill patients are often affected by systemic inflammation due to conditions such as infection, sepsis, CKD, and procedures such as cardiopulmonary bypass. These inflammatory conditions can cause increases in both serum and urinary NGAL, regardless of concurrent kidney damage. Because existing NGAL assays are unable to distinguish between distinct molecular forms of NGAL released by various organs, it is challenging to elucidate NGAL thresholds for AKI among patients with these comorbidities.

Urinary NGAL was recently approved as ProNephro AKI^™ for clinical use among pediatric ICU patients (≥3 months to <22 years) without underlying kidney disease [139]. It is intended for use within the first 24 h of ICU admission, with the goal of identifying patients who are at risk for moderate to severe AKI 48–72 h following assessment of NGAL. This test uses a clinical cutoff of 125 ng/mL of NGAL in pediatric urine, with levels <125 ng/mL interpreted as a negative result and indicative of low AKI risk within 48–72 h; alternatively, results >125 ng/mL are considered positive and indicative of an elevated risk of moderate to severe AKI (stage 2/3) within 48–72 h of testing. Availability is anticipated in 2024 and will provide clinicians with an additional tool for early evaluation and identification of AKI amongst pediatric patients.

2.10. Kidney injury molecule–1 (injury)

Kidney injury molecule–1 (KIM-1) is a 38.7-kDa type 1 transmembrane glycoprotein which mediates the phagocytosis of apoptotic cells [140,141]. It is normally expressed at low levels by healthy kidney cells, but increases in response to ischemic or toxic damage to proximal tubule cells of the kidney [140]. Ichimura et al first identified rat and human cDNAs for KIM-1. In their discovery, it was noted that KIM-1 mRNA and proteins are normally expressed at low levels in healthy kidneys but are rapidly increased in response to ischaemic kidney injury. KIM-1 expression was also observed in proliferating and dedifferentiated epithelial cells in regenerating proximal tubules, signifying that KIM-1 may also have an important role in renal cell recovery following ischaemic injury [142]. KIM-1 is one of the most promising biomarkers for detection of kidney injury because of its specificity to the proximal tubules, and the fact that it is shed into the urine in response to epithelial injury [143]. This makes it a noninvasive biomarker that is easily measured in the urine of animals and humans. Since its initial discovery, KIM-1 was among several biomarkers that were formally qualified by the PSTC and US FDA for detection of kidney injury in animal or human trial subjects [23].

Numerous preclinical studies support its utility as a specific marker of proximal tubular damage, as well as increased sensitivity in detecting tubular injury earlier than our current standard diagnostic markers [142,144–146]. Early animal toxicology studies by Vaidya et al showed that Kim-1 outperformed SCr, BUN, and urinary N-acetyl-β-D-glucosaminidase (NAG), when using histopathology as the gold standard for toxicity, a finding that was seen across multiple nephrotoxicants in a rat model. Kim-1 was also demonstrated to be a much better predictor of both low-grade (histopathology grade: 0–1) and high-grade (histopathology grade >2) injuries (area under the receiver operating characteristic curve: 0.91–0.99) when compared to SCr. In fact, SCr had very low sensitivity in detecting low-grade histopathologic damage (grade 0–1), which underscores the inability to detect early kidney injury with traditional biomarkers [147], Since then, multiple studies in rats and mice have investigated the ability of KIM-1 to detect kidney injury associated with various nephrotoxins including vancomycin, aminoglycosides, cisplatin, folic acid, contrast agents, and heavy metals [108,148–150]. In the case of nephrotoxic antimicrobials such as vancomycin and aminoglycosides, translational rat studies have shown that urinary Kim-1 is elevated within 24 h of nephrotoxin exposure, and correlates well with end-organ histopathologic damage [150,151]. Specifically for vancomycin, a number of newer studies have also correlated kidney injury response via urinary KIM-1 with real-time decreases in kidney function, via improved biomarkers such as FITC–sinistrin and iohexol [107,108].

Investigations of the performance of KIM-1 in human subjects have been limited. Han et al first reported the presence of KIM-1 elevations in human urine following ischaemic acute tubular necrosis (ATN). In this cohort, it was observed that all patients with biopsy-confirmed ATN (n = 6) had normalized urinary KIM-1 levels that were significantly higher compared to those in all other types of acute kidney failure (2.92 ± 0.61 vs. 0.63 ± 0.17, P < 0.01) [141]. Since then, KIM-1 has been investigated as a predictive marker of kidney injury in a variety of settings including patients undergoing cardiac surgery, kidney transplantation, and those with CKD [152–154]. Early results have been encouraging, with KIM-1 exhibiting promise in identifying early kidney injury related to cardio-pulmonary bypass, kidney transplantation, and among CKD patients. Despite these early results, larger, prospective studies in humans are needed. Existing studies describing the utility of KIM-1 in various clinical settings and patient populations are limited by relatively small sample sizes, single-center designs, differences in biomarkers used (i.e., KIM-1 alone or in combination with other biomarkers), and inconsistent reporting of KIM-1 levels (total amounts vs. concentration) [151–155].

Detection of KIM-1 is possible using several commercially available options, including a single-analyte, enzyme-linked immunosorbent assay (ELISA), lateral flow dipstick assay (Renastrip^™, AMSBio^®), or multiplex assay (Luminex xMAP^® and MesoScale Discovery^®). The ELISA and lateral flow assays offer the fastest way to measure KIM-1, as results can be obtained from a urine sample in 15 min. However, these assays are limited by their qualitative or semi-quantitative evaluations of KIM-1 levels, and only 1 biomarker can be assessed at a time [156]. Furthermore, in the clinical implementation of urinary kidney injury biomarkers, it is not advisable to use only 1 biomarker, due to the heterogeneity and complexities associated with kidney injury [123]. Available multiplex assays utilize either microsphere-based flow cytometry (such as Luminex xMAP^®) or patterned arrays with electrochemiluminescence detection (such as MesoScale Discovery^®) [157,158]. Both platforms offer the ability to multiplex several different protein biomarkers for a large number of samples.

Due to the sensitivity and specificity of KIM-1 in detecting specific types of kidney injury, such as vancomycin-induced nephrotoxicity, this urinary biomarker has potential as a point-of-care test. In order to establish a clinical cutoff for KIM-1, validation studies would need to be conducted in specific patient populations, in addition to development of a cost-effective point-of-care assay.

2.11. Clusterin (injury)

Clusterin is a 76- to 80-kDa secreted glycoprotein which is found in a variety of physiological fluids including plasma, urine, and cerebrospinal fluid [159]. It is intrinsically expressed at high levels during early kidney development and in response to ischaemic or toxic kidney injury in the proximal and distal tubules, glomerulus, and collecting duct [160]. Clusterin plays a role in various cellular processes such as adhesion, membrane recycling, cell cycle regulation, apoptosis, and tissue remodelling [161]. Multiple animal models have shown that clusterin expression is rapidly increased in response to drug-induced nephrotoxicity and ischaemic injury [146,162,163]. It has also been investigated as a predictor for delayed graft function after human kidney transplantation, with 1 study identifying clusterin as an early predictor for post-transplantation delayed graft function [164].

Specific to vancomycin-induced kidney injury (VIKI) in a translational rat model, urinary clusterin, along with KIM-1 and osteopontin, outperformed cystatin C and neutrophil gelatinase–associated lipocalin (NGAL) for both sensitivity and specificity. Notably, in these studies, clusterin (area under the receiver operating characteristic curve: 0.71, P < 0.001) was found to be a more sensitive predictor of early, VIKI-associated histopathologic damage at 24 h, compared to NGAL [149,150].

Recently, there was a study of pediatric patients undergoing allogeneic hematopoietic stem cell transplantation (alloHSCT), which evaluated serum and urinary clusterin, compared to KIM-1 and cysC, as alternative biomarkers of early kidney injury following alloHSCT [165]. A total of 27 children undergoing alloHSCT had serum and urine concentrations of clusterin, cysC, and KIM-1 evaluated at specific timepoints (pre-transplantation and 24 h and 1, 2, 3, and 4 weeks post alloHSCT) and compared with controls. Urinary clusterin was found to be the most sensitive biomarker for identification of early kidney injury (24 h post alloHSCT) and had an elevation and plateau similar to those of urinary KIM-1 over 4 weeks. This study provides early evidence that clusterin can be useful as a marker of subclinical and clinical AKI, specifically in the context of pediatric alloHSCT patients. Further clinical studies are still needed to validate the utility of clusterin in detecting AKI in other disease states, and in response to specific nephrotoxins.

2.12. Osteopontin (injury)

Osteopontin (OPN) is a 34-kDa phosphorylated glycoprotein that exerts both protective effects against oxidative stress and ischaemia, in addition to pro-inflammatory and profibrotic properties [161,166]. OPN is involved in several specific functions such as osteoclast regulation, tumorigenesis, macrophage accumulation, cell proliferation, and regeneration [167]. Due to its diversity in function, OPN is expressed by a variety of cells in the bone, kidney, epithelial linings, and bodily fluids such as blood and urine [168]. It also functions as a pro-inflammatory cytokine which is markedly upregulated in response to inflammation and tissue remodelling. Chronic inflammatory diseases such as Crohn disease, cancer, atherosclerosis, lupus, multiple sclerosis, and rheumatoid arthritis have all been correlated with elevated plasma OPN levels [169–173].

In the kidney, OPN is expressed in multiple areas such as the proximal and distal tubules, loop of Henle, and glomerulus [161,174]. Animal and human studies have shown that OPN is expressed by developing nephrons during the embryonic stage. In mice and humans, OPN expression is localized to the thick ascending limbs of the loop of Henle and the distal convoluted tubules. In contrast, in rats, OPN is found primarily in the descending loop of Henle [175–177]. Following damage to the kidneys, OPN expression can be significantly upregulated throughout the tubules and glomeruli. A variety of kidney conditions, including carcinomas, lupus nephritis, immunoglobulin A nephropathy, essential hypertension, glomerulonephritis, urinary stones, and allografts, have been reported to cause increased OPN expression [178].

With respect to drug-induced kidney damage, rat studies have investigated changes in expression of OPN in response to nephrotoxins. In a study of vancomycin-induced kidney injury utilizing male Sprague–Dawley rats, urinary OPN was found to perform best when kidney injury was severe (i.e., high vancomycin doses) or sustained [150]. Another study used male Wistar rats in a severe acute tubular necrosis (ATN) model with gentamicin, and found that OPN may also have a role in the regeneration of tubular epithelial cells following ATN [179]. These observed differences in the response of OPN to kidney injury may reflect differences in experimental timelines, or differences in the degree of kidney injury. Unfortunately, clinical studies supporting the utility of OPN as a kidney injury biomarker are currently lacking. Due to the differences in OPN expression that exist among different animals, disease states, and medications, future studies in humans should focus on determining whether OPN can serve as an early diagnostic marker for AKI related to medications and other disease states.

2.13. Tissue inhibitor of metalloproteinase 2 [TIMP-2] and insulin-like growth factor binding protein 7 [IGFBP7] (injury)

Tissue inhibitor of metalloproteinase 2 (TIMP-2) is a 22-kDa soluble protein that is expressed in the kidney and other tissues. TIMP-2 binds to and inhibits the activity of various metalloproteinases, thereby modulating several processes associated with cellular injury, leukocyte infiltration, and disruption of cell receptors [180–182]. Insulin-like growth factor binding protein 7 (IGFBP7) is a 26-kDa soluble protein that is also expressed in the kidney and other tissues [183]. IGFBP7 plays a role in several processes that are associated with cellular injury, including oxidative stress, inflammation, toxins, and ultraviolet radiation [184,185].

TIMP-2 and IGFBP7 are both G₁ cell cycle arrest proteins which are expressed in renal tubular cells during acute cellular stress or injury [186]. It has been shown that renal tubular cells enter a period of G₁ cell cycle arrest in response to either septic or ischaemic injury [187]. Both of these proteins play a role in this response and have been shown to be early markers for AKI related to sepsis or post surgery [188–190]. Specific to AKI, urinary TIMP-2 and IGFBP7 are expressed by the tubular cells, in response to damage or stress to the renal epithelium. The utility of using TIMP-2 and IGFBP7 in combination for early detection of AKI was first discovered in a multicenter observational study of critically ill patients at risk for AKI [186]. This study included both a discovery and validation phase for novel biomarkers of AKI, with a total of 728 patients enrolled to the biomarker validation study. Urinary TIMP-2/IGFBP7 were identified as the best-performing markers in the discovery study (AUROC = 0.77 and 0.75, respectively, for RIFLE injury or failure criteria within 12–36 h) and were noted to have additive predictive value when used together. Results are reported as an AKIRISK score, which is calculated as the product of measured concentrations of TIMP-2 and IGFBP7 (measured in nanograms per milliliter [ng/mL]), divided by 1000. Results are reported in a range of 0.04–10.0, with a cutoff of >0.3 used to identify patients at high risk for developing moderate to severe AKI within 12 h [191]. Using this cutoff, TIMP-2/IGFBP7 was found to have sensitivity between 76% and 92%, a specificity between 46% and 51%, a negative predictive value between 88% and 96%, and a positive predictive value between 27% and 31% [186].

The combination of TIMP-2/IGFBP7 was FDA approved for use in adult ICU patients in 2014 as an adjunctive assessment for moderate to severe AKI risk with 12 h of assessment. Following the FDA approval of TIMP-2/IGFBP7 as the point-of-care NephroCheck^® test, other studies have investigated its utility in detecting AKI among septic and post–cardiac surgery patients. Both sepsis and cardiac surgery are associated with high rates of AKI, around 50% and 20%, respectively [192, 193]. In several cohorts of ICU patients with sepsis-induced AKI, TIMP-2/IGFBP7 was shown to be significantly corelated with subsequent AKI incidence and severity [194–196]. Cuartero et al assessed urinary TIMP-2/IGFBP7 among a cohort of 98 septic patients in the ICU. When stratified based on the presence of AKI and the highest individual TIMP-2/IGFBP7 value measured during the first 12 h of ICU admission, TIMP-2/IGFBP7 was significantly related to AKI severity according to the AKIN criteria (P < 0.0001). The authors also found the AUROC to predict AKI of the worst TIMP-2/IGFBP7 index value was 0.798 (sensitivity 73.5%, specificity 71.4%, P < 0.0001). Index values of <0.8 obviated any need for renal replacement therapy (negative predictive value: 100%), while an index value of >0.8 predicted an AKI rate of 71% [194].

Multiple studies have also assessed the performance of TIMP-2/IGFBP7 as early markers of AKI associated with cardiac surgery. One study amongst 50 patients undergoing cardiac surgery with cardiopulmonary bypass (CPB) found that while creatinine and urine output changes took 1–3 days to develop following CPB, urinary TIMP-2/IGFBP7 began to rise as early as 4 h after CPB in patients who developed AKI [197]. More recently, an expert panel noted that TIMP-2/IGFBP7 is most commonly used in patients with sepsis, haemodynamic instability, or those undergoing surgery. Unfortunately, TIMP-2/IGFBP7 has been studied only in limited fashion for drug-induced nephrotoxicity. Existing studies in which these urinary biomarkers were used to detect cisplatin-induced AKI have yielded conflicting results, with 1 study showing no difference versus SCr, while another showed that TIMP-2/IGFBP7 predicted cisplatin-induced AKI with an AUROC of 0.92 (95% CI: 0.80–1.00) [198,199]. Current evidence supporting the utility of TIMP-2/IGFBP7 in detecting AKI among critically ill patients may be useful in decreasing nephrotoxic drug exposure among this patient population.

2.14. Biomarker limitations

An important limitation of urinary injury biomarkers that are currently under investigation is that none are completely specific for detecting AKI; moreover, depending on the type of AKI and causative agent or event, different patterns of urinary biomarker expression can be observed. Studies to date have focused primarily on the discovery, categorization, and validation of promising new biomarkers using animal models and small human studies. However, clinical implementation of these new biomarkers will require further evaluation and validation in specific patient populations and under specific kidney injury conditions. The majority of clinical studies to date have utilized healthy volunteers in the research setting, in which these biomarkers are used as adjunctive markers of kidney injury. Further use and research in other populations such as patients with different disease states, pediatric and elderly patients, and patients with differing nephrotoxin exposures is needed to fully translate these urinary injury biomarkers to everyday clinical use. Moreover, urinary injury biomarkers will likely be more informative when used as a panel. The clinical utility of urinary injury biomarkers may also be limited in settings in which 24-h urine collections are not routinely done. Existing preclinical studies have found that urinary injury biomarkers are highly sensitive and able to identify early kidney injury; however, biomarker levels are more easily interpreted as total amounts over a specified time period, as opposed to spot concentrations. In practice, urinary biomarkers are often standardized to urinary creatinine to correct for volume dilution [200,201]; however, urinary creatinine is influenced by dynamic changes in excretion with glomerular filtration rate changes. Thus, adjustment may not provide better standardization, and the best approach is not yet fully clear. Identification of blood biomarkers of kidney injury may lead to even greater clinical translational potential. Future studies of these biomarkers will be needed to evaluate clinical sensitivity and specificity, to define the time course of elevation following AKI, and to correlate with prevention of long-term outcomes such as histopathologic damage, renal failure, and/or mortality.

In conclusion, serum creatinine remains the most utilized biomarker for assessment of kidney injury (including antibiotic induced kidney injury) and function in current clinical practice. However, significant limitations related to its sensitivity and specificity have prompted recent efforts to develop newer biomarkers that are specific to the separate processes of kidney function and injury. Many of these newer methodologies show promise for detecting early injury and changes in kidney function which can have clinically significant impact on anti-infective therapy management.