Current role of biomarkers in the initiation and weaning of kidney replacement therapy in acute kidney injury

Abstract

The occurrence of acute kidney injury (AKI) in critically ill patients is often associated with increased morbidity and mortality rates. Despite extensive research, a consensus is yet to be arrived, especially regarding the optimal timing and indications for initiation of kidney replacement therapy (KRT) for critically ill patients. There is no clear guidance available on the timing of weaning from KRT. More recently, various biomarkers have produced promising prognostic prediction in such patients, regarding the need for KRT and its termination. Most of these biomarkers are indicative of kidney damage and stress, rather than recovery. However, large-scale validation studies are required to guide the cutoff values of these biomarkers among different patient cohorts so as to identify the optimum timing for KRT. This article reviews the kidney biomarkers in detail and summarizes the individual roles of biomarkers in the decision-making process for initiation and termination of the KRT among critically ill AKI patients and the supportive literature.

Key Words: Biomarker; Neutrophil gelatinase-associated lipocalin; Kidney replacement therapy; Proenkephalin 119–159; Acute kidney injury; Cystatin C

Core Tip: Routinely, > 50% of patients with acute kidney injury in intensive care require kidney replacement therapy (KRT). However, there are no clearcut guiding parameters on timing of initiation or weaning from KRT. Biomarkers have a promising role in initiation of KRT and can be used in decision-making for weaning from KRT. In this narrative review we summarize the individual roles of biomarkers and the supportive literature.

INTRODUCTION

Acute kidney injury (AKI) occurs frequently in critically ill patients and is associated with increased morbidity and mortality rates[1]. Various studies have highlighted the lack of appropriate biomarkers for both early diagnosis and progression of the condition, thus limiting their application in improving the outcomes of AKI[2]. When conservative measures fail, patients are left with no option other than undergoing kidney replacement therapy (KRT). Although the application of KRT in AKI is a crucial means of support for continuous kidney function and treating AKI-induced complications, it still poses health risks; therefore, its application should be restricted to a minimum threshold. It is crucial to understand the likelihood of kidney recovery so to avoid inadvertent and excess use of the therapy. Nowadays, extensive information and predictive tools are available to guide clinicians on possible recovery of kidney function, thus restricting the clinicians’ subjective assessment in crucial decisions regarding KRT initiation as well as termination[3].

In spite of the extensive research conducted so far, there is a lack of consensus on the ideal time to initiate KRT for critically ill AKI patients. Although early KRT initiation is justified for patients with clear indications such as severe metabolic acidosis, refractory hyperkalemia and fluid overload[4], excessive and unnecessary KRT still occurs, with accompanying complications and adverse events. Conventionally, the decision to start KRT is based on clinical assessment and conventional criteria involving urine output, serum creatinine (Cr), and hyperkalemia. However, the conventional biomarkers lack sensitivity as well as specificity for accurately predicting the need for KRT. In recent times, biomarkers have been proposed for predicting the need for KRT as well as the weaning parameters. An ideal biomarker should possess some universal attributes, such as: Ease of measurement using readily available samples (urine or blood); cost-effectiveness; and being measurable by a biologically and physiologically plausible assay with high sensitivity and specificity, with rapidly changing levels that correlate well with disease progression as well as improvement and have prognostic value. Unfortunately, none of the biomarkers that are currently used possess all these attributes. Practical limitations in biomarker-based prediction for KRT include variable cutoffs, reliance upon single measurements, and confounding by underlying comorbidities and clinical scenarios.

This review summarizes both conventional as well as novel biomarkers to guide the decision-making process for initiation and termination of KRT among critically ill AKI patients. It has been argued that the presence of AKI during intensive care unit (ICU) admission might affect the performance of biomarkers in predicting the need for KRT[5]. Table 1 provides a comprehensive overview of the clinical utilities of the biomarkers available for KRT.

Table 1 Kidney biomarkers of practical utility KRT in AKI.

Kidney biomarker	Biological role	Sample	Type of marker	Role in KRT practice	Limitations	Advantages
NGAL	At least 3 different types. Monomeric: 25-kDa glycoprotein produced by neutrophils and epithelial tissues, including tubular cells. Homodimeric: 45-kDa protein produced by neutrophils. Heterodimeric: 135-kDa protein produced by tubular cells	Urine and plasma	Damage	Predicting need for KRT with high sensitivity and specificity. Mortality in patients on KRT. Prediction of successful weaning from KRT: Not assuring	Confounding factors: Sepsis, malignancy, CKD, urinary tract infection, pancreatitis, chronic obstructive pulmonary disease, endometrial hyperplasia.	As urinary NGAL is not removed via CVVHDF, serial measurement is a real-time indicator of kidney damage
CyC	13-kDa cysteine protease inhibitor produced by nucleated human cells; freely filtered	Plasma and urine (plasma CyC: May be a marker of GFR. Urine CyC: Marker of Tubular injury)	Functional	Controversial results on the diagnostic ability for the need for KRT. Independent predictor of successful weaning from continuous KRT in AKI	Lack of specific cutoff values	Less likely to be affected by CVVHDF
PenKid	5-kDa stable fragment of endogenous opioid enkephalin	Plasma	Functional	Reliably predict AKI and need for KRT in patients with sepsis. Low pre-KRT penKid levels: Predict successful, earlier termination of KRT	Confounded by age, sex, inflammatory state, diabetes, low albumin, muscle mass, high-dose steroids	Potential to dynamically guide kidney function, prior and during ongoing KRT: Thus facilitates early and successful termination of KRT
TIMP-2 × IGFBP7	Metalloproteinases released during tubular cell cycle arrest (cell cycle arrest biomarker)	Urine	Stress	TIMP-2: More predictive of need for KRT in septic AKI patients. IGFBP-7: Performs better in surgical patients; Combined predictive ability: Better than individual biomarkers	Elevated in diabetes
suPAR	A multifaceted, glycosylphosphatidylinositol-anchored three domain protein acting as a receptor for urokinase-type plasminogen activator	Plasma	Functional	Levels at intensive care unit admission: Promising in predicting AKI progression to KRT	Since neutrophils can serve as a major source, elevated levels occur in inflammatory conditions, acute respiratory distress syndrome or different cancers. Confounding factors: CKD, polycystic kidney disease, liver disease, sepsis
FABP	Low molecular weight proteins of 14–15 kDa, belonging to lipid-binding proteins superfamily. Nine types identified: FABP-1 expressed in the proximal tubular cells; FABP-3 in the distal tubular cells	Urine and plasma	Damage	A positive prediction of KRT with use of urinary FABP-1 and FABP-3 levels	Validation of the cutoff value required. Associated with anemia in nondiabetic patients
KIM-1	Transmembrane glycoprotein produced by proximal tubular cell; released into urine after tubular cell damage; no systemic source	Urine	Damage	A good predictability for the need for KRT	Elevated in kidney cell carcinoma, chronic proteinuria, CKD, sickle cell nephropathy
Mid-regional pro-adrenomedullin	A 52 amino acid peptide with natriuretic and vasodilatory properties, and antimicrobial activity	Plasma	Damage	Prediction of requirement of KRT in children with coronavirus disease 2019 and AKI	Paucity of literature

AKI: Acute kidney injury; KRT: Kidney replacement therapy; NGAL: Neutrophil gelatinase-associated lipocalin; CKD: Chronic kidney disease; CVVHDF: Continuous veno-venous hemodiafiltration; GFR: Glomerular filtration rate; TIMP-2: Tissue inhibitor of metalloproteinase-2; IGFBP7: Insulin-like growth factor-binding protein 7; suPAR: Soluble urokinase plasminogen activator receptor; FABP: Fatty acid-binding proteins; KIM-1: Kidney injury molecule 1; PenKid: Proenkephalin 119–159.

BIOMARKERS FOR INITIATION OF KRT

Table 2 summarizes the major trials on biomarkers in initiating KRT for AKI patients[5-16]. The biomarkers that are routinely used are briefly described.

Table 2 Summary of studies on role of biomarkers for initiation of kidney replacement therapy.

Ref.	Biomarker studied	Type of study	No. of patients	AUROC (95%CI)	Cutoff	Sensitivity (95%CI)	Specificity (95%CI)	Positive predictive value (95%CI)	Negative predictive value (95%CI)
Klein et al[5]	Urine output	Meta-analysis	604	0.614 (0.389-0.840)
Klein et al[5]	Fractional excretion of sodium	Meta-analysis	240	0.718 (0.619-0.816)
Klein et al[5]	Blood urea nitrogen	Meta-analysis	283	0.732 (0.661-0.802)
Klein et al[5]	Plasma/serum Cr	Meta-analysis	2969	0.764 (0.732-0.796)
Hjortrup et al[6]	Plasma Cr	Prospective	222	0.74 (0.67–0.82)	166	63	75	36	90
Pipili et al[7]	UNGAL	Prospective	310	0.727	106.7			66	86
Klein et al[5]	UNGAL	Meta-analysis	3195	0.720 (0.638-0.803)
Klein et al[5]	PNGAL/SNGAL/whole-blood NGAL	Meta-analysis	4391	0.755 (0.706–0.803)
Tiranathanagul et al[8]	PNGAL	Prospective	47	0.813 (0.66-0.90)	960	72.2	9.6	81.25	83.8
Tiranathanagul et al[8]	UNGAL	Prospective	47	0.806 (0.63-0.98)	2600	54.5	90.9
Haase et al[9]	PNGAL/UNGAL	Meta-analysis	2538	0.782 (0.648-0.917)	278.3 (141.9-381.6)	76.0 (65.1-84.4)	80.3 (59.5-91.9)
Cruz et al[10]	PNGAL	Prospective	301	0.82 (0.70-0.95)	150	0.87 (0.60–0.98)	0.65 (0.60–0.71)	0.12 (0.06–0.19)	0.99 (0.96–1.00)
Cemil et al[11]	PNGAL	Prospective	60	0.84 (0.74-0.94)	615	82	80
Nisula et al[12]	UNGAL	Prospective	1042	0.839 (0.797-0.880)	449	83.1	78.5	30	90
Hjortrup et al[6]	PNGAL	Prospective	211	0.70 (0.61-0.78)	641	69	64	28	88
Hjortrup et al[6]	UNGAL	Prospective	162	0.62 (0.51-0.73)	641	46	77
Hanson et al[13]	UNGAL	Prospectivecohort–acute falciparum malaria	163	0.68 (0.55-0.80)	1832
Lukasz et al[14]	SNGAL	Prospective Cohort-Shiga toxin	39	0.76 (0.61-0.91)	330	83.3	60	76.9	69.2
Senthilkumaran et al[15]	PNGAL	Prospective and Russell’s Viper bite	309	1.0	493.75	98	100	79	98
Albert et al[16]	UNGAL	Meta-analysis	5454	0.74 (0.66–0.82)	83	78	67

AUROC: Area under the receiver operating characteristic curve; CI: Confidence interval; Cr: Creatinine; PNGAL: Plasma neutrophil gelatinase-associated lipocalin; SNGAL: Serum neutrophil gelatinase-associated lipocalin; UNGAL: Urinary neutrophil gelatinase-associated lipocalin.

TRADITIONAL BIOMARKERS

Urine output

Urine output is a useful clinical marker for estimating glomerular filtration rate (GFR), although its specificity remains unclear. Patients with AKI may be classified as anuric, oliguric (< 100 mL and < 400 mL per 24 h, respectively) or polyuric based on their urine output. A meta-analysis conducted earlier, involving 604 patients [pooled area under the curve (AUC) of 0.614 (0.389–0.840)][5], concluded its limited role in predicting the need for KRT. AKI guidelines and many other studies also recommend that spontaneous improvement in urine output from oliguric to nonoliguric state (> 400 mL/d) may be taken into account as a reliable clinical parameter to consider weaning from KRT[7,17].

Serum Cr

Serum Cr is the most common and widely used parameter for laboratory diagnostics to detect AKI. However, this measure is influenced by a wide range of parameters such as age, gender, muscle wasting, fluid overload and drugs. Serum Cr levels start increasing only after 50% of the nephrons are injured. This drawback may delay the diagnosis of AKI and worsen patient outcomes. Thus, it can be considered that serum Cr is a marker for kidney function instead of injury[18].

In the meta-analysis comprising 15 studies with 2969 patients, a pooled AUC of 0.7645 was found for serum Cr, with a low predictive role[7]. Two individual studies[8,19] also found lower AUC values for using serum Cr as a trigger for KRT initiation. Although high Cr levels intensify the consideration for initiating KRT, a specific cutoff value at which KRT is to be initiated remains unclear. Thus, serum Cr can be identified as a poor, late and insensitive biomarker for detection of AKI and eventually, the need for KRT. A similar pattern can be observed for using blood urea nitrogen (BUN) values to initiate KRT in AKI patients[8,19]. The guidelines recommend urine Cr clearance, with > 15–20 mL/min, as a reliable clinical parameter to consider weaning from KRT[17]. Although the traditionally considered parameters cannot predict the need for KRT in AKI patients individually, it is possible that integrated plasma Cr, serum sodium, BUN values, albuminuria, and clinical data of fluid balance and daily urine output could predict the need for KRT with 88% accuracy in noncritically ill patients. However, the clear cutoff values to predict the timing for initiating KRT in AKI patients is yet to be defined in a precise manner[19].

Furosemide stress test

Furosemide stress test (FST) is an excretory functional assessment of the kidneys. Recently, it has been proposed as a promising method to predict kidney disease, the need for KRT in patients with AKI[20] and complications after kidney transplantation[21]. Nonresponders to FST have a greater need for KRT than responders.

NOVEL BIOMARKERS

Neutrophil gelatinase-associated lipocalin

Neutrophil gelatinase-associated lipocalin (NGAL), a 25-kDa glycoprotein, is present at low concentrations in neutrophil granules in organs such as the kidneys, stomach, colon, trachea and lungs. It is excreted via the urine owing to its low molecular weight and can be detected easily[22]. NGAL is resistant to fragmentation; therefore, it is a useful noninvasive biomarker for early detection of AKI[23,24]. However, it is secreted from kidney tubular cells, hepatocytes and immune cells under various disease conditions. For instance, epithelial injury in lungs, colon and predominantly the kidneys can induce the secretion of NGAL. Exposure of the kidneys to different toxins and ischemia can also cause increased expression of NGAL in kidney cortical tubules, and these increased levels can be easily detected in blood and urine.

Two systematic reviews found a strong association between plasma and urine NGAL that demands initiation of KRT[9,25]. A prospective observational study conducted among 301 patients found that plasma NGAL is a strong predictor for KRT initiation within five subsequent days of diagnosis (AUC = 0.82)[10]. Likewise, a small study conducted among 47 patients found that plasma NGAL alone (cutoff = 960 ng/mL, sensitivity 72.2% and specificity 89.6%, and positive and negative predictive values of 81.25% and 83.8%, respectively) or in combination with APACHE II score can predict the need for KRT in early stages of AKI[8].

A study conducted among 60 patients admitted to the emergency department in a Turkish hospital[11] concluded that NGAL is a useful biomarker for predicting the need for emergency hemodialysis. The NGAL cutoff value of 615 ng/mL had a sensitivity of 82%, specificity of 80%, and AUC of 0.84 [95% confidence interval (CI): 0.74–0.94; P < 0.001] regarding the prediction of emergency hemodialysis. In another study conducted among 1042 patients, 83 of whom received KRT, an association was found with the highest value of urine NGAL during the first 24 h (AUC = 0.839, 95%CI: 0.797–0.880) with the best cutoff value of 449 (95%CI: 219–538) ng/mL[20]. A prospective study found that the prognostic capacity of urine NGAL for initiating KRT was maximum at 10 days after sample collection, with an AUC under the receiver operating characteristic curve (AUROC) of 0.727[7].

Contradicting the findings discussed above, the predictive ability of urine and plasma NGAL (AUROC = 0.62 and 0.70, respectively) was lowest among a subset of AKI patients with severe sepsis, in a multicenter trial conducted in Denmark[6]. Septic AKI has been proposed to be a secondary phenomenon next to apoptosis of the kidney cells rather than necrosis of the tubular cells. This phenomenon may explain the poor predictive value of the NGAL biomarker among this group of patients. A similar finding, i.e., poor predictive ability of urinary NGAL was found among patients with severe acute falciparum malaria[13]. However, serum NGAL has consistently showed a strong association with the need for KRT in patients with AKI, caused by Shiga toxin-producing Escherichia coli infection-associated hemolytic uremic syndrome[14], and among a large cohort of victims of Russell’s viper bite[15]. Recently, two meta-analyses and systematic reviews found AUC values of 0.80 and 0.82 for plasma NGAL and AUC values of 0.75 and 0.84 for urinary NGAL in predicting the need for KRT[16,26]. Thus, NGAL can be highly recommended as a useful biomarker for predicting the need for KRT initiation, although a specific cutoff value needs to be defined before its clinical application.

Cystatin C

Cystatin C (CyC), a 13.3-kDa low molecular weight protein, is filtered freely through the glomerular membrane, and plasma CyC levels are considered to be a surrogate for GFR. CyC is produced by all the cells in the body and is released into the circulation at a constant rate. In healthy individuals, all the filtered CyC is completely reabsorbed by the proximal kidney tubules with almost nil secretion. This phenomenon makes the CyC levels in urine undetectable. The presence of CyC in urine corresponds to kidney tubular damage; therefore, CyC is a potential biomarker for kidney damage[27].

The literature shows contradictory results for the diagnostic ability of serum CyC for predicting the need for KRT. The predictive role of serum and urine CyC for KRT was limited in an initial study among 80 patients[28]. Another study found a moderate-to-good predictive value for initiation of KRT, with an increase in plasma CyC level ≥ 50% during AKI–RIFLE risk stage[29]. One study[30] showed good diagnostic accuracy of serum CyC levels for KRT on the day of ICU admission, while a contradictory study[7] found that serum CyC was inferior to serum Cr in predicting the need for KRT. Recently, a meta-analysis conducted with 1079 patients concluded that both plasma and urinary CyC can provide good prediction about the need for KRT, with a pooled AUC of 0.768 (95%CI: 0.729–0.807)[5].

Proenkephalin 119-159

Proenkephalin 119–159 (penKid), a highly stable fragment of the endogenous opioid enkephalin, plays a significant role in the regulation of the kidney functions. PenKid has been used as a biomarker to measure kidney function in acute, unstable and critical conditions[31]. It can be used as an alternative to measure GFR, as levels are unaffected by inflammation, age, gender or hemodynamic fluctuations[32]. High penKid levels can be relied upon to predict AKI and the need for KRT in patients with sepsis[33].

Tissue inhibitor of metalloproteinase-2/Insulin-like growth factor-binding protein 7

Tissue inhibitor of metalloproteinase-2 (TIMP-2) × Insulin-like growth factor-binding protein 7 (IGFBP7) are involved in G1 cell cycle arrest that occurs during the early stages of renal tubular cell injury[34]. TIMP-2 is highly predictive of the need for KRT in septic AKI patients, whereas IGFBP7 provides better prediction for surgical patients[34]. The combined predictive ability (AUC > 0.8) and consistent outcomes are better than for the individual biomarkers. In a prospective study conducted among 100 patients with sepsis-induced AKI[35], a stepwise increase in TIMP-2 × IGFBP7 concentration was observed with stage-by-stage worsening of AKI. TIMP-2 × IGFBP7 concentration continues to remain elevated over the first 48 h in patients who are in need of KRT. Eventually, its concentration declines over the first 12–48 h in patients who do not require KRT. These phenomena emphasize the need for sequential measurements to predict the progression in patients with septic AKI[35].

Recently, a systematic review conducted by Klein et al[5] found that TIMP-2 × IGFBP7 concentration was the best urinary biomarker for predicting the need for KRT (pooled AUC = 0.86, 95%CI: 0.79–0.93). However, this review included only four small studies; all of which were originally designed to predict AKI. Koyner et al[36] found inferior performance of TIMP-2 × IGFBP7 with an AUC of 0.61. This was attributed to the inclusion of mild AKI patients in the study with only a single measurement of TIMP-2 × IGFBP7. Similar findings were observed in postoperative patients, developing AKI after open or endovascular repair of thoracoabdominal aortic aneurysm. In this case, when the TIMP2 × IGFBP7 levels were measured 6–12 h after the surgical procedure, it successfully predicted the need for temporary KRT, with a sensitivity of 55.56%, specificity of 90.91% and AUC of 0.694[37].

Soluble urokinase-type plasminogen activator receptor

The concentration of serum soluble urokinase-type plasminogen activator receptor (suPAR) is elevated in critically ill patients during their first week of admission and is considered to be a predictor of mortality[38]. Its concentration is predictive of a decline in GFR among chronic kidney disease patients. These values show the promising performance of suPAR in accurately predicting the progression of AKI to KRT. The highest baseline levels of suPAR were found in patients who eventually needed KRT. However, the patients with AKI stage 2/3 had lower levels and recovered without KRT[35].

Recently, another study found significantly higher suPAR levels in patients who required KRT[39]. The suPAR levels can classify AKI into transient AKI, persistent AKI, and AKI that requires KRT, at any point of time within 7 days after enrolment in the study. Those with suPAR levels > 12.7 ng/mL were at higher risk for KRT or even death, with an adjusted odds ratio of 7.48 (95%CI: 3.0–18.63). In a retrospective cohort study conducted among postoperative cardiac patients, the doubling of preoperative suPAR value had an odds ratio of 1.92 for patients who needed KRT[40].

Fatty acid binding protein

Fatty acid binding proteins (FABPs) are low molecular weight proteins weighing between 14 and 15 kDa, distributed in different types of tissues, belonging to a superfamily of lipid-binding proteins. A total of nine types of FABPs have been identified so far. FABP-1 is a liver-type FABP that is highly expressed in hepatocytes, whereas FABP-3 is found mainly in heart muscle[41,42]. Both isoforms are also found in the kidney tubules, with FABP-1 being expressed predominantly in the proximal tubular cells and FABP-3 in the distal tubular cells[43]. The urinary FABP-1 levels are often associated with the risk of AKI progression with an estimated sensitivity of 74.5% and a specificity of 77.6%[44]. The initial studies that determined the association between FABP-1 levels and the need for KRT in AKI revealed contradictory results; i.e., a negative correlation was found by Nakamura et al[45] while a positive correlation was found by Ferguson et al[46].

Another study by Dihazi et al[47] found a positive prediction between KRT and the utilization of both urinary FABP-1 and FABP-3 levels, after normalizing for urine protein levels, with high specificity and sensitivity values (FABP-1: 95%CI, cutoff > 0.2895; 88%–100% sensitivity; 88%–100% specificity; FABP-3: 95%CI, cutoff > 0.04008; 92%–100% sensitivity; 88%–99% specificity). Both urinary FABP-1 and FABP-3 possess the ability to best predict the need for KRT (AUC of 0.9995 and 1, respectively) in a meta-analysis that evaluated 13 biomarkers[5]. Both FABP-1 and FABP-3 have shown promising results in terms of predicting the need for KRT in AKI patients, with the need for larger studies to validate the defined cutoff values.

Kidney injury molecule-1

The presence of kidney injury molecule-1 (KIM-1) in urine is highly specific for kidney injury. However, only one study evaluated its use in critically ill patients, in whom KIM-1 showed good predictability for the need for KRT and 7 day mortality in patients with AKI[48]. The role of KIM-1 as a biomarker has a promising future not only in diagnosis but also in therapeutics.

Mid-regional pro-adrenomedullin

The use of mid-regional pro-adrenomedullin showed promising results in predicting the requirement for KRT in children with COVID-19 and AKI, among a cohort of 64 patients (AUC = 0.69)[49].

BIOMARKERS FOR WEANING FROM KRT

The weaning from KRT usually marks the beginning of kidney function recovery and gradual improvement in urine output is the most commonly used indicator for weaning off KRT[17,50]. However, there is a lack of precise tools to predict the time for weaning from KRT. Timely and early discontinuation of KRT can reduce the cost of needless prolonged periods of KRT and the associated complications, and allow for better resource allocation[51]. Some of the urinary biomarkers that correlate with GFR recovery, at the time of weaning from KRT, include urinary NGAL, KIM-1, hemojuvelin, C-C motif chemokine ligand 14 (CCL14), interleukin-18 (IL-18), and FABP-1. Table 3 summarizes the studies conducted on the role played by biomarkers in weaning from KRT[52-58].

Table 3 Summary of studies on role of biomarkers for weaning of KRT.

Ref.	Biomarker studied	Type of study	No. of patients	AUROC (95%CI)	Cutoff	Sensitivity (95%CI)	Specificity (95%CI)
Chen et al[52]	SNGAL	Prospective	110	0.81	403	91	61
Chen et al[52]	SNGAL	All AKI, prospective nonseptic AKI	70	0.89	403	81	89
Chen et al[52]	SNGAL	Prospective and septic AKI	40	0.65	782	63	67
Komaru et al[53]	UNGAL	Prospective	105	0.81 (0.71-0.88)	186	94	60
Kim et al[54]	SNGAL	Prospective	110	0.654 (0.539-0.768)	-	90.5	44.9
von Groote et al[55]	PenKid–pre-KRT	Post hoc analysis	210	0.582 (0.505–0.660)	95.25	50	67
von Groote et al[55]	PenKid at day 10	Post hoc analysis	210	0.681 (0.594–0.767)	77.1	60	71
von Groote et al[55]	PenKid at day 28	Post hoc analysis	210	0.660 (0.570–0.750)	74.9	52	75
Kim et al[54]	Serum CyC	Prospective	110	0.739 (0.624-0.853)	-	76.2	62.9
Yang et al[56]	Serum CyC	Prospective	69	0.525 (0.408-0.643)	-	-	-
Aregger et al[57]	Kynurenic acid on day 1	Prospective	92	0.72 (0.59-0.85)	-	-	-
Aregger et al[57]	Kynurenic acid on day 2	Prospective	92	0.80 (0.69-0.93)	-	-	-
Aregger et al[57]	Kynurenic acid on day 3	Prospective	92	0.78 (0.66-0.90)	-	-	-
Pan et al[58]	Liver-fatty acid-binding proteins/creatinine	Prospective	140	0.79	-	-	-

AUROC: Area under the receiver operating characteristic curve; CI: Confidence interval; KRT: Kidney replacement therapy; SNGAL: Serum neutrophil gelatinase-associated lipocalin; AKI: Acute kidney injury; UNGAL: Urinary neutrophil gelatinase-associated lipocalin; CyC: Cystatin C; PenKid: Proenkephalin 119–159

NGAL

An earlier study involving small sample size[52], using NGAL for predicting successful weaning from KRT did not yield promising results. Serum NGAL levels, when initiating KRT, were high among the nonsurvivors, although no specific association was found between serum NGAL levels and kidney function recovery[59]. However, a recent multicenter prospective study conducted in Japan among 133 patients identified that the lower levels of urinary NGAL (optimal cutoff value 186 ng/mL; 94% sensitivity; and 60% specificity; AUROC 0.81) at continuous venovenous hemodiafiltration (CVVHDF) discontinuation, have an association with successful weaning from the KRT (93.8 ng/mL vs 999 ng/mL, P < 0.01). The findings were also reflected in the subgroup analysis of the patients with sepsis[53].

Urinary NGAL is not removed via CVVHDF; therefore, the serial measurements can function as a real-time indicator in ICU patients with kidney damage. A rising NGAL value corresponds to worse outcomes[60]. A prospective multicenter study evaluated the usefulness of restarting KRT in patients, based on their higher urinary NGAL levels on day 2 after discontinuation of continuous renal replacement therapy (CRRT) for expected recovery[61]. However, the multivariate analysis outcomes inferred that restarting KRT had no association with NGAL levels, whereas only the incremental Cr levels had an independent correlation with restarting KRT[61].

PenKid

PenKid has the potential to dynamically guide kidney function recovery prior to and during KRT, thus facilitating its early and successful termination. A post hoc analysis of the landmark ELAIN study revealed an association between successful and earlier termination of KRT in patients with low pre-KRT penKid levels of < 89 pmol/L[55].

CyC

A prospective observational trial involving 110 patients with a need for CRRT found that the serum cystatin level, measured during cessation of CRRT, was an independent predictor of successful weaning from KRT in AKI patients (AUC = 0.739)[54]. Similar findings were found in another study in which serum CyC, measured at the time of KRT discontinuation, predicted the recovery of kidney function at day 60[56]. As opposed to serum NGAL, CyC is less likely to be affected by CVVHDF; therefore, it helps in monitoring residual kidney function. It is also highly accurate in predicting the kidney outcomes during CRRT[62,63].

Other biomarkers

Higher plasma C-terminal fibroblast growth factor-23 levels, collected when initiating KRT, were found to predict lower probability of weaning the AKI survivors from dialysis, with levels > 2050 RU/mL predicting mortality[64]. A prospective study showed that urinary L-FABP/Cr (AUC = 0.79), with a cutoff < (log) level 2.2 μg/g Cr, had the highest predictive ability of weaning from dialysis for > 90 d[58].

ADVANCEMENTS IN BIOMARKER RESEARCH BEYOND THE ROUTINE BIOMARKERS

microRNAs

microRNAs (miRNAs) are a group of noncoding RNAs involved in the pathogenesis of AKI that remain stable in the bodily fluids and are easily detected. Preclinical studies have already established several miRNAs (e.g., miR-16-5p, miR-24-3p, miR-101-3p, miR-127-3p, miR-26-5p and miR-126-3p) with promising applications for future use, not only in the early diagnosis of AKI with an AUC of 1, but also as a therapeutic tool, named theranostics[65].

Mitochondrial DNA

A close association has been found between mitochondrial DNA and loss of kidney function, and it has been proposed as a marker of kidney injury[66].

Exosomal biomarkers

Urinary exosomes are cell-derived extravesicular particles that are secreted from various cells in the urinary system. Urinary exosomes possess few advantages over routine biomarkers such as stable structure, lower susceptibility to urinary factors, early changes in its content and noninvasive acquisition[67]. Urinary exosomes hold much potential in diagnosing AKI, the need for KRT, monitoring graft rejection and drug-induced nephrotoxicity.

Proteomic biomarkers

Proteomic technology has revolutionized medical research in terms of identification, quantification and functional characterization of the proteins and peptides in biological samples. The entire set of proteins expressed in a biological sample can be fully characterized with the help of proteomics. Potential biological pathways are associated with kidney remodeling, repair and regeneration; therefore, high throughput proteomics that detects serum protein levels in AKI has potential for predicting the need for KRT. This combination of multiple peptide biomarkers also increases stability and accuracy[68].

Metabolomics

The metabolomics of kidney diseases including components of the kynurenine pathway, tryptophan metabolism and the intermediates in the urea cycle has been extensively investigated. High urinary excretion of kynurenic acid, a key inflammatory metabolite of the tryptophan degradation pathway, was associated with delayed recovery from AKI and longer need for KRT[57]. Other metabolomics used to identify the biomarkers for KRT includes urinary metabolites such as 1-methylnicotinamide, glycine and lactate[69].

Wearable biosensors

Wearable biomedical monitoring systems allow continuous measurement of the critical biomarkers to monitor kidney disease and overall health of the patient. However, cohort validation studies and performance evaluation are needed to underpin their clinical utility, especially under acute conditions[70]. The routine clinical application of these techniques still has a long way to go as the protocols required for the standardization of these methods are still at a nascent stage. These potential diagnostic tools should be validated against the current conventional biomarker assays, including urinary albumin, serum Cr and CyC. Such a comparative assessment helps in understanding whether the novel proteomic and metabolomic biomarkers provide any improved diagnostic accuracy at reasonable costs and lesser time.

APPLICATION OF COMBINATION OF BIOMARKERS

A panel of biomarkers can provide better insights into AKI diagnosis and yield results from different pathophysiological mechanisms of the AKI subtypes, with an emphasis on the most affected segments of the renal tubules. Various studies and guidelines have stressed the importance of combining two or more biomarkers or a combination of biomarkers and biochemical tests, to predict the need for KRT to achieve accurate results.

A prospective study conducted among 106 mechanically ventilated patients with AKI, with 50 patients requiring KRT within 7 days of ICU admission, found that urinary NGAL, serum CyC and serum Cr were appropriate independent predictors for KRT requirement during ICU admission (P < 0.0001)[7]. For patients who presented without AKI on admission (n = 85), a combination of serum CyC and APACHE II score was found to best estimate the time for initiating KRT[28]. TIMP-2 × IGFBP7 concentration at 24 h and suPAR with CyC provided an additional predictive value that outperformed the other individual tests in terms of predicting KRT[35]. Those patients who required KRT showed the highest level of TIMP-2 × IGFBP7 and suPAR, against the low median levels of TIMP-2 × IGFBP7 (< 0.3 ng/mL over 48 h) and the lowest suPAR levels at baseline levels among patients with stable AKI grade 1. Thus, it is better to combine TIMP-2 × IGFBP7 concentration and suPAR with CyC to achieve the best performance in order to predict the need for KRT.

Another study that combined high urinary biomarker CCL14 level and negative FST achieved significantly higher predictive value for the development of KRT indications, with an AUC of 0.87 (95%CI: 0.82–0.92) compared with FST or CCL14 alone (AUC = 0.79; 95%CI: 0.74–0.85 and AUC = 0.83; 95%CI: 0.77–0.89; P < 0.001, respectively)[71]. The secondary analysis study conducted earlier involving the DAMAGE (Dublin Acute Biomarker Group Evaluation) protocol to assess the utility of a panel of 14 urinary biomarkers showed that in patients with AKI stage 1 or 2 within 48 h after ICU admission, eight biomarkers were associated with progression towards worse AKI, KRT, or even death in 7 d[72]. The best predictors for these outcomes included CyC, IL-18, albumin and NGAL. When the patients who developed AKI within the first 7 days after enrolment were evaluated, IL-18, NGAL, albumin and monocyte chemotactic protein-1 levels were independently associated with KRT or death within 30 days.

CONCLUSION

Numerous kidney biomarkers have been studied extensively in both blood and urine for multiple cohorts. However, no single biomarker has been validated so far in large sample-size studies that can predict the need for KRT and determine the appropriate timing for its initiation and termination[73]. Despite the efforts taken to identify an optimal biomarker for the past three decades, their usefulness at the bedside is largely limited. Internationally, weak recommendations have been proposed for the inclusion of biomarker data with clinical assessments for identifying the patients who may need KRT, facilitating an optimal timing for KRT initiation and predicting the duration and recovery of AKI and KRT[74].

The biomarkers, which predict the need for KRT, are the markers of kidney damage/injury and the loss of glomerular function. Until now, an increase in the TIMP-2 × IGFBP7 concentration or suPAR values has been set as a threshold for initiating KRT in critically ill patients with AKI. Of all the currently available biomarkers, NGAL can be highly recommended for predicting the need for KRT initiation, but a specific cutoff value needs to be defined before its clinical application. However, the superiority of these novel biomarkers over standard parameters such as urine output and Cr is yet to be conclusively confirmed through large-scale empirical research. Although these parameters predict the likelihood of the need for KRT, their usefulness to predict the optimum timing for initiating KRT to achieve the best outcomes is yet to be clinically proven. The decision to start the KRT is based on the level of an individual biomarker or the level of kidney injury, and involves consideration of multiple patient factors such as comorbidities, severity of acute disease, fluid balance and metabolic integrity.

PenKid has been identified as a promising biomarker in a landmark post hoc analysis of the ELAIN trial, to guide timely and successful weaning from KRT. It is the first step towards the implementation of precision medicine for critically ill patients. There is a need to conduct large-scale validation studies that can investigate the correlation among various biomarkers and their threshold values, to the outcomes of AKI, to allow timely prediction for initiating and weaning from KRT.