What are the most effective real-time markers of kidney cell damage?

NGAL, [TIMP‑2]·[IGFBP7], and KIM-1 are the most effective real-time markers of kidney cell damage, showing high diagnostic accuracy when measured at appropriate time windows and achieving superior performance when combined in multi-marker panels.

Abstract

Real‐time markers for kidney cell damage most consistently report high diagnostic accuracy when measured in the appropriate clinical context and time window. Urinary neutrophil gelatinase‐associated lipocalin (NGAL) rises within 2–6 hours after injury, with studies reporting area under the curve (AUC) values up to 0.91, sensitivities of 84–100%, and specificities as high as 99.5%. Urinary [TIMP‑2]·[IGFBP7] is typically measured at 6–12 hours post‐injury and shows AUC values ranging from 0.70 to 0.94, with reported sensitivities of 42–96% and specificities up to 95%. Urinary kidney injury molecule‑1 (KIM‑1) measured at 12–24 hours post‐insult yields moderate performance (AUC 0.66–0.84) with sensitivities between 75.9% and 91.6% and specificities up to 95.2%.

When combined in multi‐marker panels, these biomarkers often achieve superior performance. For example, panels including NGAL, [TIMP‑2]·[IGFBP7], and cystatin C have produced AUC values as high as 0.98, while combinations with additional markers have reached similar levels of discrimination. These findings, drawn from a diverse set of populations and clinical settings—ranging from cardiac surgery and intensive care to emergency presentations—support the use of NGAL, [TIMP‑2]·[IGFBP7], and KIM‑1 as effective real‐time indicators of kidney cell damage.

Methods

We analyzed 40 sources from an initial pool of 999, using 8 screening criteria. Each paper was reviewed for 5 key aspects that mattered most to the research question. More on methods

Papers identified with Elicit search
n = 999
Papers screened using: Real-time Biomarker Detection, Human Participants, Diagnostic Performance Assessment, Appropriate Study Design, Comparative Assessment, Clinical Study Setting, Primary Research Quality, Clinical Utility Assessment
n = 999

Papers screened out
n = 959
Papers included for extraction
n = 40

Paper search

Using your research question “What are the most effective real-time markers of kidney cell damage?”, we searched across over 126 million academic papers from the Semantic Scholar corpus. We retrieved the 999 papers most relevant to the query.

Screening

We screened in sources based on their abstracts that met these criteria:

• Real-time Biomarker Detection: Does this study evaluate biomarkers that can detect kidney cell damage in real-time or near real-time (within 24 hours maximum)?
• Human Participants: Does this study involve human participants with actual or potential kidney cell damage?
• Diagnostic Performance Assessment: Does this study assess diagnostic performance metrics (such as sensitivity, specificity, positive/negative predictive values, AUC, or diagnostic accuracy)?
• Appropriate Study Design: Is this study a prospective cohort study, cross-sectional diagnostic accuracy study, randomized controlled trial, systematic review, or meta-analysis?
• Comparative Assessment: Does this study compare novel biomarkers to established reference standards or other biomarkers?
• Clinical Study Setting: Is this study conducted in a clinical setting (not exclusively animal studies, in vitro studies, or cell culture studies)?
• Primary Research Quality: Is this study a primary research study or high-quality secondary research (not a case report, case series, editorial, or opinion piece)?
• Clinical Utility Assessment: Does this study assess the diagnostic performance or clinical utility of biomarkers (not just measure biomarker levels without performance evaluation)?

We considered all screening questions together and made a holistic judgement about whether to screen in each paper.

Data extraction

We asked a large language model to extract each data column below from each paper. We gave the model the extraction instructions shown below for each column.

• Study design: Specify the exact type of study design used. Look in the methods section for precise description. Possible types include:

  • Cross-sectional study
  • Prospective multicenter study
  • Case-control study
  • Cohort study If multiple design elements are present, list all. If design is not clearly stated, write “design not clearly specified”. Be as specific as possible, including any unique methodological features of the study design.

• Participant characteristics: Extract the following details:

  • Total number of participants
  • Patient population (e.g., critically ill patients, cardiac surgery patients)
  • Age range or mean age
  • Gender distribution
  • Specific inclusion/exclusion criteria If any of these details are missing, note “not reported”. Use exact numbers and percentages from the text. If ranges are provided, include both minimum and maximum values.

• Biomarkers investigated: List ALL biomarkers examined in the study for detecting kidney cell damage. Include:

  • Full name of biomarker
  • Source of biomarker (e.g., urine, serum)
  • Measurement method (if specified) If multiple biomarkers were studied, list them in order of appearance in the text. If a specific combination or interaction of biomarkers was studied (e.g., [TIMP-2]·[IGFBP7]), note this explicitly.

• Biomarker performance metrics: Extract quantitative performance data for biomarkers, specifically:

  • Area Under the Curve (AUC)
  • Sensitivity
  • Specificity
  • Confidence intervals (if reported)
  • Cutoff values used Ensure you capture the specific context of performance (e.g., detecting early vs. late stage kidney injury). If multiple time points or conditions are reported, extract data for each.

• Primary outcome definition: Describe how acute kidney injury (AKI) was defined and diagnosed in the study. Look for:

  • Specific diagnostic criteria used
  • Staging system (e.g., KDIGO, RIFLE)
  • Specific thresholds (e.g., 50% increase in serum creatinine)
  • Method of clinical adjudication If multiple definitions or methods are used, list all. If the definition is not clearly stated, write “outcome definition not clearly specified”.

Results

Characteristics of Included Studies

Summary of Study Characteristics:

• Most commonly evaluated biomarkers:
  • Neutrophil gelatinase-associated lipocalin (NGAL): 29 studies
  • Kidney injury molecule-1 (KIM-1): 21 studies
  • [TIMP-2]·[IGFBP7]: 12 studies
  • Interleukin-18 (IL-18): 11 studies
  • Liver-type fatty acid-binding protein (L-FABP): 8 studies
  • Cystatin C (CysC): 14 studies
  • Other biomarkers (e.g., NAG, B2M, SCr/creatinine, albumin, clusterin, GSTs, HGF, microRNAs, etc.): each evaluated in 1–4 studies
• Study populations:
  • Critically ill: 7 studies
  • ICU: 6 studies
  • Cardiac surgery: 7 studies
  • ED/ER: 6 studies
  • Pediatric: 2 studies
  • Transplant: 3 studies
  • Drug-induced/nephrotoxic settings: 5 studies
  • Hospitalized (general): 2 studies
  • Other populations (diabetes, heart failure, poisoning, snake envenoming, neonates, AAA surgery, at-risk adults, etc.): 1 study each
• AKI definitions used:
  • KDIGO: 15 studies
  • AKIN: 8 studies
  • RIFLE: 3 studies
  • SCr increase (not otherwise specified): 3 studies
  • GFR-based: 1 study
  • No mention found of a clearly specified AKI definition: 13 studies
• Primary outcomes:
  • Early AKI detection/diagnosis: 19 studies
  • Prediction (including risk stratification, prediction of AKI, or progression): 13 studies
  • Prognosis (renal recovery, adverse events, delayed graft function): 6 studies
  • Drug-induced AKI detection: 5 studies
  • Other outcomes (biomarker timing/performance, diabetic nephropathy, etc.): 3 studies
• Heterogeneity: There was no consistent AKI definition across all studies. Several studies evaluated multiple biomarkers or included multiple populations or outcomes.

Effects

Individual Biomarker Performance

Summary of Diagnostic Performance:

• Area Under the Curve (AUC):
  • 5 biomarkers had minimum AUC values in the poor range (<0.7)
  • 8 biomarkers had AUC values in the fair range (0.7–0.8)
  • 8 biomarkers had AUC values in the good range (0.8–0.9)
  • 5 biomarkers had maximum AUC values in the excellent range (≥0.9)
• Sensitivity:
  • 2 biomarkers had minimum sensitivity values <50%
  • 2 biomarkers had sensitivity values in the 50–69% range
  • 7 biomarkers had sensitivity values in the 70–89% range
  • 5 biomarkers had maximum sensitivity values ≥90%
  • We didn’t find sensitivity data for albumin, peptide signature, or clusterin
• Specificity:
  • 5 biomarkers had minimum specificity values in the 50–69% range
  • 7 biomarkers had specificity values in the 70–89% range
  • 7 biomarkers had maximum specificity values ≥90%
  • We didn’t find specificity data for albumin, peptide signature, or clusterin
• No mention found of biomarkers with minimum specificity values <50% in the included studies.

Combined Biomarker Performance

Several studies reported that combinations of biomarkers outperformed individual markers. Key findings from these studies include:

• Dong et al., 2017: Combination of NGAL, IL-18, and [TIMP-2] yielded an AUC of 0.973 at 12 hours post-cardiopulmonary bypass, higher than any single marker.
• Che et al., 2010: Combination of five biomarkers (cystatin C, NGAL, IL-18, retinol-binding protein, NAG) achieved an AUC of 0.98.
• Gupta et al., 2025: Combination of [TIMP-2]·[IGFBP7], NGAL, and cystatin C yielded an AUC of 0.92, which improved to 0.94 when clinical risk factors were included.
• Elmedany et al., 2017: Combination of NGAL, KIM-1, and urine sediment yielded an AUC of 0.906.

Summary:

• In the studies where this was evaluated, multi-marker panels and combinations were reported to outperform single biomarkers, particularly for early detection and risk stratification.
• The most effective combinations typically included [TIMP-2]·[IGFBP7] and NGAL, sometimes with cystatin C or clinical variables.

Timing of Detection and Clinical Context

Summary of Timing and Context:

• Biomarker types: Three individual biomarkers (NGAL, [TIMP-2]·[IGFBP7], KIM-1), one multi-marker panel, and one urine sediment analysis were evaluated for timing.
• Optimal detection windows:
  • NGAL: 2–6 hours post-injury
  • [TIMP-2]·[IGFBP7]: 6–12 hours post-injury
  • KIM-1: 12–24 hours post-injury
  • Multi-marker panel: 6–24 hours post-injury
  • Urine sediment: 2–48 hours post-injury
• Performance metrics:
  • AUC values for all five biomarkers/panels ranged from 0.66–0.98.
  • NGAL: AUC >0.9 (early)
  • [TIMP-2]·[IGFBP7]: AUC 0.70–0.94
  • KIM-1: AUC 0.66–0.84
  • Multi-marker panel: AUC 0.906–0.98
  • Urine sediment: AUC 0.79–0.865
  • Sensitivity values were found for NGAL (84–100%), [TIMP-2]·[IGFBP7] (89–96%), and KIM-1 (75.9–91.6%).
  • Specificity values were found for NGAL (80–99.5%), [TIMP-2]·[IGFBP7] (50–95%), KIM-1 (61.8–95.2%), and urine sediment (91–98.6%).

Factors Affecting Biomarker Performance

• Patient population: Performance varied by clinical context, including ICU, surgery, sepsis, drug-induced injury, pediatric, oncology, and neonates.
• Comorbidities: Some markers, such as [TIMP-2]·[IGFBP7], were reported to be robust across comorbidities (chronic kidney disease, congestive heart failure, diabetes), while others may be confounded.
• Timing: Early sampling was critical. NGAL typically rose first, followed by [TIMP-2]·[IGFBP7], KIM-1, and others.
• AKI definition: Variability in AKI definitions (KDIGO, AKIN, RIFLE, expert adjudication) affected comparability across studies.
• Analytical methods: Differences in assay platforms, normalization (e.g., to creatinine), and cutoff values impacted reported performance.
• Clinical variables: Integration with clinical risk factors and scoring systems improved predictive accuracy.

Summary:

• The effectiveness of real-time kidney cell damage markers was influenced by clinical context, timing, comorbidities, and methodological factors.
• Standardization of definitions, timing, and analytical methods is needed to optimize and compare biomarker performance across studies.