Abstract
Clinicians have access to limited tools that predict which patients with early AKI will progress to more severe stages. In early AKI, urine output after a furosemide stress test (FST), which involves intravenous administration of furosemide (1.0 or 1.5 mg/kg), can predict the development of stage 3 AKI. We measured several AKI biomarkers in our previously published cohort of 77 patients with early AKI who received an FST and evaluated the ability of FST urine output and biomarkers to predict the development of stage 3 AKI (n=25 [32.5%]), receipt of RRT (n=11 [14.2%]), or inpatient mortality (n=16 [20.7%]). With an area under the curve (AUC)±SEM of 0.87±0.09 (P<0.0001), 2-hour urine output after FST was significantly better than each urinary biomarker tested in predicting progression to stage 3 (P<0.05). FST urine output was the only biomarker to significantly predict RRT (0.86±0.08; P=0.001). Regardless of the end point, combining FST urine output with individual biomarkers using logistic regression did not significantly improve risk stratification (ΔAUC, P>0.10 for all). When FST urine output was assessed in patients with increased biomarker levels, the AUC for progression to stage 3 improved to 0.90±0.06 and the AUC for receipt of RRT improved to 0.91±0.08. Overall, in the setting of early AKI, FST urine output outperformed biochemical biomarkers for prediction of progressive AKI, need for RRT, and inpatient mortality. Using a FST in patients with increased biomarker levels improves risk stratification, although further research is needed.
Keywords: acute renal failure, clinical nephrology, diuretics
AKI is the most common reason for inpatient nephrology consultation and carries an increased risk for morbidity and mortality.1–3 Despite a dramatic increase in the incidence of AKI over the last decade, physicians still lack the clinical tools to determine the likelihood of AKI progression (defined as a worsening of AKI stage, such as progressing from stage 1 to stage 2 or 3) for those with early AKI.4 Improving patient risk stratification will be crucial as therapeutic trials aim to enroll patients with early AKI who are at highest risk for RRT or inpatient death.5 In addition, the past decade has seen an explosion of studies seeking to discover new biomarkers of AKI.6,7 In patients with early AKI (e.g., stage 1), several biomarkers of AKI, including plasma neutrophil gelatinase-associated lipocalin (NGAL), urinary IL-18, tissue inhibitor of metalloproteinases (TIMP-2) and IGF-binding protein-7 (IGFBP-7) have demonstrated variable ability to predict AKI progression.8–12 However, despite intense investigation, the utility of these and other biomarkers remains unclear, and most nephrologists and intensivists do not have clinical access to these assays.
We recently demonstrated that the 2-hour urine output after a standardized high-dose furosemide stress test (FST, 1 mg/kg of furosemide in naive patients or 1.5 mg/kg in those with prior exposure) in clinically euvolemic patients with early AKI has the predictive capacity to identify those with severe and progressive AKI.13 The area under the receiver-operating characteristic curve (AUC) for the urine output 2 hours after FST to predict progression to AKI Network (AKIN) Stage-3 AKI in 77 patients (±SEM) was 0.87±0.09; P=0.001. The ideal cutoff for predicting progressive AKI during these first 2 hours was a urine volume <200 ml (100 ml/hr) with a sensitivity of 87.1% and a specificity of 84.1%.13
In the current study we compare the performance of several biomarkers of AKI, including the fractional excretion of sodium (FeNa), urine and plasma NGAL, urine albumin-to-creatinine ratio, urinary IL-18, kidney injury molecule-1 (KIM-1), TIMP2, IGFBP-7, and uromodulin, with that of FST for the prediction of several clinical end points, including progressive AKI, need for RRT, and inpatient mortality.
Results
As previously reported, 25 of 72 (32.5%) patients progressed to Acute Kidney Injury Network (AKIN) stage 3 AKI after evaluation with FST.13 Table 1 lists the clinical characteristics of those with and without progressive AKI. Patients with progressive AKI were more likely to have cirrhosis, had lower pre-furosemide urinary flow rates, had evidence of more severe tubular injury before FST (according to the George Washington University–Urinalysis Scoring System [GWU-USS]), and were more likely to have AKIN stage 2 at the time of FST.
Table 1.
Patient characteristics
| Variable | Entire Cohort (n=77) | Nonprogressors (n=52) | AKIN Stage 3 Progressors (n=25) | P Value |
|---|---|---|---|---|
| Demographics | ||||
| Age (yr) | 65.3±1.6 | 63.8±2.2 | 68.2±1.9 | 0.13 |
| Median weight (IQR) (kg) | 82.0 (68.5–96.8) | 80.0 (68.9–94.9) | 85.0 (68.3–99.8) | 0.64 |
| Men (%) | 42.8 | 36.5 | 56 | 0.14 |
| Race, n (%) | 0.20 | |||
| African American | 44 (57.1) | 29 (55.6) | 15.0 (60) | |
| White | 23 (29.9) | 15 (28.8) | 8 (32) | |
| Hispanic | 10 (13.0) | 8 (15.4) | 2 (8) | |
| Comorbidities, n (%) | ||||
| CKD | 24 (31) | 17 (32.7) | 7 (28.0) | 0.80 |
| Hypertension | 60 (78) | 41 (78.8) | 19 (76) | 0.78 |
| Cirrhosis | 4 (5) | 0 (0) | 4 (16) | 0.01 |
| Congestive heart failure | 25 (32.5) | 15 (29) | 10 (40) | 0.44 |
| Diabetes mellitus | 35 (44) | 22 (41.5) | 13 (52) | 0.47 |
| Nephrotoxic exposures, n (%) | ||||
| NSAIDs | 8 (10) | 6 (2) | 2 (1) | 0.99 |
| Aminoglycosides | 1 (1) | 0 (0) | 1 (4.0) | 0.63 |
| Amphotericin | 2 (3) | 2 (4) | 0 (0) | 0.99 |
| Contrast medium | 21 (27) | 15 (28.8) | 6 (23.1) | 0.79 |
| Cardiac surgery | 9 (11.7) | 6 (11.5) | 3 (12.0) | 0.99 |
| Sepsis | 15 (19.5) | 12 (23.1) | 3 (12.0) | 0.36 |
| Clinical data | ||||
| Baseline eGFR (ml/min per 1.73 m2) | 68.6±4.1 | 60.0±8.8 | 73.3±4.2 | 0.15 |
| Baseline urine flow rate (ml/hr) | 74.6±11.6 | 95.7±16.3 | 29.7±4.2 | 0.001 |
| Furosemide naive, n (%) | 29 (37.7) | 23 (44.2) | 6 (24) | 0.13 |
| Median furosemide dose (IQR) (mg) | 100 (80–131.5) | 100 (73.3–132.3) | 100 (87.5–135) | 0.35 |
| Urine cast score | 2.3±0.13 | 2.1±0.16 | 2.7±0.23 | 0.05 |
| Median FeNa (IQR) (%) | 0.58 (0.14–1.1) | 0.60 (0.13–1.4) | 0.51 (0.13–1.0) | 0.96 |
| CV SOFA score | 1.16±0.03 | 1.05±0.2 | 1.5±0.4 | 0.37 |
| APACHE II score | 17.8±1.11 | 16.5±1.2 | 21.6±2.5 | 0.08 |
| AKIN stage at enrollment, n (%) | ||||
| Stage 1 (total) | 41 (53.2) | 34 (65.4) | 7 (28.0) | |
| Urine output criteria | 15 (19.4) | 11 (21.1) | 4 (16.0) | |
| Serum creatinine criteria | 38 (49.4) | 32 (61.5) | 6 (24.0) | |
| Stage 2 (total) | 36 (46.7) | 18 (34.6) | 18 (72.0) | 0.003 |
| Urine output criteria | 26 (33.8) | 11 (21.2) | 15 (60.0) | |
| Serum creatinine criteria | 15 (28.8) | 10 (19.2) | 5 (20.0) | |
| Outcomes | ||||
| Death | 16 (20.7) | 7 (13.4) | 9 (36.0) | 0.04 |
| AKIN stage 3 | 25 (32.4) | NA | 25 (100) | NA |
| RRT | 11 (14.2) | NA | 11 (44.0) | NA |
| Death/AKIN | 32 (41.5) | 7 (13.4) | 25 (100) | 0.001 |
Data presented as mean±SEM unless otherwise noted. NSAID, nonsteroidal anti-inflammatory drug; CV SOFA, Cardiovascular Sequential Organ Failure Assessment; APACHE, Acute Physiology and Chronic Health Evaluation; NA, not applicable.
Prediction of Progression to AKIN Stage 3
The 2-hour urine output (UOP) after FST was associated with the development of AKIN stage 3 AKI with an AUC (±SEM) of 0.87±0.05 (P<0.001) (Table 2). Several biomarkers were also able to significantly predict progression to stage 3, but no individual biomarker provided an AUC>0.80. Plasma NGAL performed the best, with an AUC of 0.75±0.08 (P=0.007). When compared head to head, the AUC of FST was significantly higher than that of all other biomarkers with the exception of plasma NGAL (Table 2). Combining individual biomarkers with the 2-hour UOP demonstrated that some biomarkers improved the risk prediction of AKIN stage 3, while other biomarkers diminished the FST’s ability to predict AKI progression (Table 2). However, despite increases in the AUC, including an AUC with IGFBP7×TIMP2+2-hour UOP of 0.90±0.06 (P=0.20), the FST alone and the biomarkers combined with FST did not significantly differ.
Table 2.
AUCs for prediction of progression to AKI stage 3
| Biomarker | AUC±SEM | P Value for Biomarker Alone | P Value Compared With FST alone | AUC of Biomarker and FST±SEM | P Value for Biomarker and FST Compared With FST Alone |
|---|---|---|---|---|---|
| FST (2-hr UOP) | 0.87±0.05 | <0.001 | NA | NA | NA |
| Urine NGAL | 0.65±0.06 | 0.04 | 0.002 | 0.84±0.05 | 0.10 |
| Urine IL-18 | 0.65±0.07 | 0.04 | 0.009 | 0.85±0.05 | 0.89 |
| Urine KIM-1 | 0.63±0.06 | 0.07 | 0.007 | 0.86±0.05 | 0.79 |
| Uromodulin | 0.54±0.07 | 0.54 | 0.002 | 0.85±0.05 | 0.94 |
| Urine IGFBP-7 | 0.62±0.09 | 0.20 | <0.001 | 0.88±0.05 | 0.57 |
| Urine TIMP-2 | 0.70±0.08 | 0.03 | 0.02 | 0.83±0.06 | 0.20 |
| Urine IGFBP-7×TIMP-2 | 0.69±0.08 | 0.04 | 0.01 | 0.90±0.06 | 0.35 |
| Urine Creatinine | 0.48±0.08 | 0.77 | <0.001 | 0.84±0.06 | 0.85 |
| Urine ACR | 0.56±0.07 | 0.45 | 0.002 | 0.84±0.06 | 0.32 |
| FeNa | 0.51±0.07 | 0.92 | <0.001 | 0.83±0.06 | 0.47 |
| Plasma NGAL | 0.75±0.08 | 0.007 | 0.10 | 0.86±0.07 | 0.53 |
NA, not applicable; ACR, albumin-to-creatinine ratio.
Prediction of Receipt of RRT
Eleven patients (14.2% of the total cohort) required RRT following FST (Table 1). Table 3 demonstrates the AUCs for the FST and biomarkers to predict the need for RRT. Two-hour urine output outperformed all individual biomarkers, providing an AUC of 0.86±0.08 (P<0.001), with no single biomarker providing a statistically significant P value or an AUC>0.65. FST significantly outperformed several of the individual biomarkers, including urine NGAL, IL-18, KIM-1, and IGFBP-7 for predicting the receipt of RRT. Combining an individual biochemical biomarker with the FST did not significantly increase the AUC for the prediction of RRT above that seen with FST alone; several biomarkers combinations decreased the AUC compared with the AUC for FST alone (Table 3).
Table 3.
AUCs for prediction of receipt of inpatient RRT
| Biomarker | AUC±SEM | P Value for Biomarker Alone | P Value Compared With FST alone | AUC of Biomarker and FST±SEM | P Value for Biomarker and FST Compared With FST Alone |
|---|---|---|---|---|---|
| FST (2-hr UOP) | 0.86±0.08 | 0.0001 | NA | NA | NA |
| Urine NGAL | 0.50±0.08 | 0.96 | 0.0006 | 0.88±0.06 | 0.35 |
| Urine IL-18 | 0.61±0.07 | 0.26 | 0.03 | 0.85±0.09 | 0.70 |
| Urine KIM-1 | 0.61±0.10 | 0.27 | 0.05 | 0.85±0.09 | 0.77 |
| Uromodulin | 0.55±0.11 | 0.60 | 0.07 | 0.89±0.06 | 0.65 |
| Urine IGFBP-7 | 0.57±0.12 | 0.61 | 0.05 | 0.90±0.06 | 0.29 |
| Urine TIMP-2 | 0.62±0.12 | 0.33 | 0.17 | 0.83±0.09 | 0.57 |
| Urine IGFBP-7×TIMP-2 | 0.61±0.13 | 0.37 | 0.10 | 0.89±0.07 | 0.23 |
| Urine creatinine | 0.64±0.11 | 0.19 | 0.24 | 0.84±0.09 | 0.90 |
| Urine ACR | 0.67±0.09 | 0.13 | 0.28 | 0.86±0.08 | 0.51 |
| FeNa | 0.64±0.09 | 0.18 | 0.11 | 0.85±0.09 | 0.27 |
| Plasma NGAL | 0.52±0.13 | 0.88 | 0.07 | 0.80±0.13 | 0.92 |
NA, not applicable; ACR, albumin-to-creatinine ratio.
Prediction of Inpatient Death
Sixteen (20.7%) of the entire cohort died in the hospital. Those with AKIN stage 3 (n=9 [36%]) were more likely to die than nonprogressors (n=7 [13.4%]; P=0.04) (Table 1). Two-hour UOP after FST provided an AUC of 0.70±0.09 (P=0.02) for inpatient death; FST and KIM-1 were the only biomarkers displaying significant relationships with this end point. Despite a higher AUC, FST was only significantly better than FeNa at forecasting inpatient death (Table 4). Combining biomarkers with FST provided no significant improvement over FST in mortality prediction.
Table 4.
AUCs for prediction of inpatient mortality
| Biomarker | AUC±SEM | P Value for Biomarker Alone | P Value Compared With FST alone | AUC of Biomarker and FST±SEM | P Value for Biomarker and FST Compared With FST Alone |
|---|---|---|---|---|---|
| FST (2-hr UOP) | 0.70±0.09 | 0.02 | NA | NA | NA |
| Urine NGAL | 0.66±0.08 | 0.06 | 0.82 | 0.71±0.08 | 0.64 |
| Urine IL-18 | 0.57±0.09 | 0.40 | 0.30 | 0.63±0.09 | 0.55 |
| Urine KIM-1 | 0.68±0.07 | 0.04 | 0.90 | 0.70±0.08 | 0.81 |
| Uromodulin | 0.52±0.08 | 0.79 | 0.18 | 0.65±0.08 | 0.71 |
| Urine IGFBP-7 | 0.65±0.11 | 0.19 | 0.48 | 0.66±0.11 | 0.45 |
| Urine TIMP-2 | 0.58±0.12 | 0.47 | 0.78 | 0.62±0.12 | 0.60 |
| Urine IGFBP-7×TIMP-2 | 0.64±0.11 | 0.22 | 0.46 | 0.66±0.12 | 0.36 |
| Urine creatinine | 0.42±0.07 | 0.37 | 0.38 | 0.65±0.07 | 0.75 |
| Urine ACR | 0.47±0.09 | 0.68 | 0.10 | 0.61±0.10 | 0.46 |
| FeNa | 0.41±0.08 | 0.31 | 0.04 | 0.66±0.08 | 0.76 |
| Plasma NGAL | 0.43±0.11 | 0.55 | 0.34 | 0.56±0.11 | 0.94 |
NA, not applicable; ACR, albumin-to-creatinine ratio.
Prediction of Composite Endpoint of Death or AKIN Stage 3
Thirty-two patients (41.5% of the total cohort) met the composite end point of death or progression to AKIN stage 3. The 2-hour UOP significantly predicted this composite end point, with an AUC of 0.81±0.06 (P<0.001) (Table 5). Plasma and urine NGAL, urine KIM-1, and IGFBP7×TIMP-2 also significantly predicted death or progression (AUC range, 0.64–0.69). FST provided a significant increase in AUC for the prediction of this end point compared with IL-18, KIM-1, uromodulin, and urine creatinine. While combining uromodulin and FST increased the AUC to 0.85±0.06, there was no significant difference between FST and biomarker combinations (Table 5).
Table 5.
Prediction of the composite of AKIN stage 3 and death
| Biomarker | AUC±SEM | P Value for Biomarker Alone | P Value Compared With FST alone | AUC of Biomarker and FST±SEM | P Value for Biomarker and FST Compared With FST Alone |
|---|---|---|---|---|---|
| FST (2-hr UOP) | 0.81±0.06 | <0.0001 | NA | NA | NA |
| Urine NGAL | 0.69±0.06 | 0.006 | 0.07 | 0.82±0.06 | 0.89 |
| Urine IL-18 | 0.63±0.07 | 0.07 | 0.009 | 0.82±0.06 | 0.87 |
| Urine KIM-1 | 0.64±0.06 | 0.04 | 0.04 | 0.82±0.06 | 0.81 |
| Uromodulin | 0.54±0.07 | 0.58 | 0.004 | 0.85±0.06 | 0.31 |
| Urine IGFBP-7 | 0.65±0.08 | 0.07 | 0.19 | 0.79±0.08 | 0.80 |
| Urine TIMP-2 | 0.66±0.08 | 0.06 | 0.18 | 0.80±0.08 | 0.75 |
| Urine IGFBP-7×TIMP-2 | 0.68±0.08 | 0.03 | 0.27 | 0.78±0.08 | 0.93 |
| Urine Creatinine | 0.54±0.07 | 0.56 | 0.007 | 0.83±0.06 | 0.23 |
| Urine ACR | 0.50±0.07 | 0.96 | 0.002 | 0.82±0.06 | 0.32 |
| FeNa | 0.49±0.07 | 0.84 | 0.009 | 0.80±0.06 | 0.31 |
| Plasma NGAL | 0.69±0.08 | 0.03 | 0.27 | 0.80±0.08 | 0.76 |
NA, not applicable; ACR, albumin-to-creatinine ratio.
Prediction of Outcomes Using Biomarker Cutoffs
Forty-four (57.1%) patients had a urine NGAL concentration >150 ng/ml before FST. In this group, FST significantly predicted the development of all four outcomes (progression to stage 3, receipt of RRT, inpatient death, and the composite end point). The AUC for the prediction of RRT was 0.91±0.06 (P<0.001), while the AUC for the composite end point of stage 3 AKI or death was 0.89±0.06 (P<0.001) (Table 6). There were 33 patients (41.5%) with a pre-FST TIMP-2×IGFBP-7 concentration >0.3. In this subset of patients, FST provided an increased AUC for the prediction of progression to stage 3 AKI (0.90±0.06; P<0.001) and receipt of RRT (0.91±0.08; P<0.001) (Table 6).
Table 6.
Prediction of patient outcomes by FST in patients with elevated biomarker levels
| Variable | Urine NGAL>150 ng/ml (n=44) | Urine TIMP-2×IGFBP-7>0.3 (n=32) | ||||
|---|---|---|---|---|---|---|
| Patients, n (%) | AUC±SEM | P Value | Patients, n | AUC±SEM | P Value | |
| Progression to AKIN stage 3 | 19 (43.2) | 0.86±(0.06 | <0.001 | 11 (34.4) | 0.90±0.06 | <0.001 |
| Receipt of inpatient RRT | 7 (15.9) | 0.91±0.06 | <0.001 | 4 (12.5) | 0.91±0.08 | 0.009 |
| Inpatient death | 12 (27.3) | 0.72±(0.10 | 0.03 | 5 (15.6) | 0.53±0.19 | 0.85 |
| Progression to AKIN stage 3 or inpatient death | 24 (54.5) | 0.89±0.06 | <0.001 | 14 (43.8) | 0.81±0.10 | 0.003 |
Discussion
We compared the ability of urine output after a standardized furosemide challenge to several contemporary AKI biomarkers, measured just before furosemide administration, to predict several clinical outcomes in 77 patients with early AKI. Our data demonstrate that urine output in the first 2 hours after FST outperforms several biomarkers of AKI for the prediction of AKI progression and future receipt of RRT. Specifically, FST was significantly better than our complete panel of urinary biomarkers at predicting progression to AKIN stage 3. The addition of biomarkers to FST results did not provide any additional benefit. Similarly, FST outperformed all other biomarkers in predicting the end point of receipt of RRT and inpatient death.
Interestingly, when we used prespecified biomarker cutoffs and assessed the performance of FST to predict patient outcomes in a “high-risk” subset, we saw increases in the AUC compared with FST alone. These analyses were modeled on clinical care of patients with myocardial infarction and pulmonary embolus in which a noninvasive tissue injury biomarker, such as serum troponin or d-dimer, is measured to assist in risk stratification and assess the need for more invasive testing (e.g., cardiac stress test, computed tomography, or ventilation-perfusion scanning for pulmonary embolus). While the number of these “high-risk” patients was low, these analyses demonstrate a potential use for the selective combination of biochemical biomarkers and FST (Table 6). The improved risk stratification achieved from biomarker cutoffs is in contrast to the analyses in which biomarker data from all patients is used (Tables 2–5) and merits further investigation in larger prospective cohorts.
To our knowledge, this study is the first to compare a diuretic challenge and biomarkers in patients with AKI. However, our investigation builds on existing literature that similarly demonstrates response to diuretics in the setting of decompensated heart failure predicts adverse patient outcomes.14 While physicians have been informally using some form of the FST in a variety of clinical settings for decades, our standardization of this renal stress test has permitted a formal evaluation of the outcomes and comparisons with biomarkers. Importantly, before receiving the FST all patients in this study were clinically evaluated as euvolemic and stable for furosemide challenge, as determined by their treating physicians, before receiving this intervention. In addition, intravenous fluids to provide volume replacement for FST-induced urinary losses were permissible for euvolemic patients at the discretion of the treating team (see Concise Methods).13
Our data demonstrate the promise of FST in improving risk stratification of patients with early AKI. Previously, several biomarker studies investigated the utility of biomarkers of AKI to predict AKI progression in those with the earliest stages of AKI.8–11 The Translational Research Investigating Biomarker Endpoints in AKI (TRIBE AKI) published biomarker data on 380 adults who developed AKIN stage 1 AKI after cardiac surgery.9 Among these patients with AKI defined by changes in serum creatinine, 45 (11.8%) developed progressive AKI and 10 (22%) died during the postoperative hospital stay. TRIBE AKI demonstrated that albumin-to-creatinine ratio, urinary IL-18, and plasma NGAL predicted progression to AKIN stage 3 after adjustment for clinical factors known to be associated with severe AKI (AUC, ±SEM 0.78±0.04, 0.77±0.04, and 0.8±0.04, respectively). Using this same clinical model, a plasma NGAL>322 ng/ml predicted a 7.7-fold increased risk of AKI progression. Similarly, the Southern AKI Network (SAKInet) measured 32 candidate biomarkers in the urine of 95 adults who developed AKIN stage 1. In this study, 23 patients (24.2%) developed worsened AKI or died while 13 (13.7%) met the secondary end point of AKIN stage 3 or death. In the similarly sized study, urinary IL-18 provided an AUC of 0.89 (95% confidence interval, 0.75 to 0.95) for this secondary end point, with IL-6, cystatin C, KIM-1, and NGAL all providing AUCs >0.80.11 Similarly, Kashani et al. recently demonstrated that in a cohort of 744 critically ill patients (mixed medical and surgical intensive care unit [ICU]), some with Kidney Disease Improving Global Outcomes (KDIGO) stage 1 AKI and some with no AKI, TIMP2×IGFBP-7 measured early in the ICU course (at study enrollment) predicted who would develop KDIGO stage 2 or 3 AKI within the subsequent 12 hours (AUC, 0.80). When these biomarkers were analyzed separately for AKI severity, IGFBP-7 and TIMP-2 exhibited an AUC of 0.76 and 0.79, respectively.10
Of note, differences between these studies and our pilot investigation may explain the discrepancy in biochemical biomarker performance. Our pilot cohort contained patients with AKIN stage 2 AKI. While we attempted to preferentially enroll patients who had AKIN stage 1, the consent process for most patients meant they were enrolled and received the FST 6–12 hours after the clinical criteria for stage 1 had been established. By that time several patients had already progressed to stage 2. This heterogeneity of AKI severity affects biomarker performance.15,16 The analyses from the TRIBE AKI,9 SAKInet,11 and Kashani10 et al. cohorts all excluded patients with stage 2 AKI at the time of biomarker measurement. In addition to the slightly increased severity of AKI in our cohort, the timing of our biomarker measurements was slightly different from that in the TRIBE-AKI and SAKInet protocols. Given the time lag involved in identifying and consenting patients for the FST protocol, biomarkers were often measured 6–12 hours after clinical evidence of AKI was determined. Thus, it is not surprising that injury or structural biomarkers did not perform well in this cohort because the kinetics of these biomarkers are not suited for detection at this delayed timepoint in this established AKI population.6,17,18 Despite the poor performance of some biochemical biomarkers in our pilot data, the aforementioned studies provide ample evidence to support the continued investigation of biomarkers at the time of AKI for the prediction of progression, as well as a role for investigation of biomarkers of renal recovery. We expect that when FST is investigated in a larger cohort and is used in combination with biomarkers of AKI, these indicators of renal tubule integrity will synergize and further improve risk stratification for AKI severity and other adverse patient outcomes.
As clinicians move into an era of AKI prophylactic and therapeutic trials we need safe, reliable and standardized methods to predict AKI progression.5,19–21 Currently we lack the tools to reliably determine which patients with early AKI (or no AKI) will progress to severe outcomes, such as the need for RRT or death. In our study 32.5% of those with early AKI progressed to stage 3 AKI while 27.3% received RRT or died. Thus, for every 3–4 patients we enrolled, only 1 experienced a hard clinical end point. In an AKI interventional trial a screening tool such as FST (or FST combined with biomarkers) may provide investigators resource and financial efficiency. Maximizing enrollment of patients with AKI who will complete a clinically meaningful end point (most severe stages of AKI, receipt of dialysis, or inpatient death) will be important for clinical trialists, and the FST is potentially an optimal tool for this task.22
Our study has several strengths. We relied on standardized AKI staging criteria (AKIN) that are internationally accepted and can be readily duplicated in future investigations. Similarly, we used hard clinical endpoints, such as receipt of RRT or inpatient death, that can also be duplicated. We have compared the FST to eight of the most widely investigated AKI biomarkers, including many of those that have shown promise for the prediction of AKI progression.9–11 In addition, those running the biologic assays were blinded to patient outcomes.
However, given the size of our pilot cohort and the limited number of patient events, we are not powered to extend our multivariate modeling beyond the simple investigations of FST-biomarker combinations presented. Thus, our ability to analyze the combinations of biochemical biomarkers, without FST, to forecast the development of adverse patient outcomes was limited. Given the time-lag that is inherent to conducting clinical research in the setting of clinical illness, however, measuring damage or injury biomarkers 6–12 hours after clinical AKI may not be an appropriate comparison. Moreover, extrapolation of our data to multivariable models, as presented in our original pilot data, would not be statistically appropriate given our cohort size and the number of biomarkers we investigated. However, we have previously shown that FST remains a significant predictor of AKI progression even after controlling for pre-FST urine output, GWU-USS cast score, change in serum creatinine, and prior furosemide exposure.13 Finally, we are limited by the lack of follow-up after discharge for those who survived their FST index hospitalization.
In summary, in patients with early AKI stable enough to undergo FST, 2-hour urine output serves as a promising tool for assessment of AKI severity and prognosis. As a functional biomarker of AKI, FST was superior to the panel of biochemical biomarkers for all end points. Improving risk prediction in those with early AKI is likely to alter patient care and clinical decision making, as well as facilitate enrollment into future therapeutic AKI trials. Our pilot study did not demonstrate statistically improved risk prediction when FST was combined with biochemical biomarkers of AKI. This may have been a limitation of our sample size and the timing of biomarker measurement. Given the wealth of data supporting the potential utility of biomarkers of AKI, larger prospective validations of the FST, with biomarkers, should be conducted.
Concise Methods
As previously described,13 we assembled two separate cohorts of critically ill patients with stage 1 or 2 AKIN criteria11,23–25 who were given a standardized dose of furosemide and had their urinary response and inpatient outcomes assessed. Briefly, our study combines patients who were part of the SAKInet cohort at George Washington University and were retrospectively identified as having received an FST,11,23,24 and patients who had their FST prospectively from June 2009 through December 2012 at George Washington University (NCT00673244) or the University of Chicago (NCT01275729).
Study Criteria
To receive the FST, patients had to meet the following inclusion criteria: (1) age older than 18 years and admission in an ICU; (2) AKIN stage 1 (6 hours of oliguria [<0.5 ml/kg per hour] or 0.3-mg/dl rise in serum creatinine or 50%–100% increase in serum creatinine above baseline) or AKIN stage 2 (12 hours of oliguria [<0.5 ml/kg per hour] or 100%–200% increase in serum creatinine above baseline during a 48-hour period); (3) indwelling bladder catheter; (4) presence of granular or epithelial cell casts on urine sediment (defined by GWU-USS score ≥2)26 or a FeNa > 1.0%; and (5) opinion of the treating clinical team that patient was well resuscitated and sufficiently clinically stable for the intervention.
Patients were excluded if they met any of the following exclusion criteria: (1) baseline eGFR<30 ml/min per 1.73 m2, (2) history of renal allograft, (3) known pregnancy, (4) evidence of obstructive uropathy, (5) evidence of active bleeding, (6) allergy or known sensitivity to loop diuretics, (7) achievement of AKIN stage 3 criteria, or (8) evidence of volume depletion at the time of furosemide administration and (9) prior episode of AKIN stage 1 or 2 during the same hospital admission.
The study procedures for the administration of the FST (furosemide dose of 1.0 mg/kg or 1.5 mg/kg) and the option for intravenous fluid replacement matching urine output in the 6 hours after FST have all been previously described.13 All ICU patients were being screened for eligibility on a daily basis. Once a patient was identified, the investigator research team approached the clinical treating service to determine whether the patient was hemodynamically appropriate and sufficiently clinically stable to undergo an FST. Inherent in the assessment for FST is that the patient was not undergoing active volume expansion or active resuscitation. In addition, for patients who had previously been resuscitated, readiness for FST implied that the patient was no longer hypovolemic or under-resuscitated. The treating team had the final determination regarding the patient’s ability to receive the FST and was presented with the option of real-time replacement of urinary losses with isotonic intravenous fluid. Once the treating team approved, the patient (or proxy) was approached for informed consent. Informed consent was often obtained 6–12 hours after the patient met AKIN stage 1 criteria. This time delay explains how patients with AKIN stage 2 were enrolled. Additionally, if a patient was not naive to furosemide (or other diuretic), the FST was not administered until at least 6 hours after the prior diuretic dose.
We defined the baseline creatinine as the nadir outpatient serum creatinine from the 6 months before the admission. If no serum creatinine values from before the admission were available, then the baseline serum creatinine was defined as the lowest serum creatinine from this admission.
Outcomes
The primary outcome was the progression to AKIN stage III (need for RRT, increase in serum creatinine of 300% over baseline, a serum creatinine level>4.0 mg/dl with an acute rise of at least 0.5 mg/dl, urine output of 0.3 ml/kg per hour×24 hours or anuria for 12 hours) within 14 days of FST. The secondary outcomes included inpatient mortality and the receipt of RRT.
Biomarker Assays
All biomarker samples and measures were collected just before the administration of furosemide. TIMP-2 and IGFBP-7 were measured by Astute Medical Inc. as previously described.10,12 Plasma and urine NGAL were measured using a commercially available ELISA (Bioporto, Hellerup, Denmark) in the Biomarker Lab in the Section of Nephrology and Hypertension at the Cincinnati Children’s Hospital. Urinary IL-18 was measured at Cincinnati Children’s Hospital using an ELISA (MBL International, Woburn, MA) as previously described.9,17,27 Urine creatinine concentration was measured using the Jaffe method on a Siemens Dimension Xpand Plus HM clinical analyzer. Urinary KIM-1 was measured at Cincinnati Children’s Hospital using a previously described ELISA (intra-assay coefficient of variation, 5.5%; interassay coefficient of variation, 5.0%) (Covance, Central Laboratory Services, Indianapolis, IN).28 Uromodulin was measured at Cincinnati Children’s Hospital using an ELISA (MD Biosciences, St. Paul, MN) (intra-assay coefficient of variation, 8.5%; interassay coefficient of variation, 11%). Urine albumin and sodium assays were measured by immunoturbidimetry on a Siemens Dimension Plus with a Heterogeneous Immunoassay clinical analyzer (Siemens Healthcare Diagnostics, Deerfield, Ill), as previously described.29 All laboratory personnel were blinded to patient outcomes during the measurement of the biomarkers.
Statistical Analyses
We assessed the distribution of demographic and clinical variables. Differences between proportions of patients with certain characteristics were assessed with chi-squared tests, Fisher exact test, t-tests, and Mann–Whitney test as appropriate. The primary analysis was to compare the ability of urine output response to the FST and biomarkers to predict the primary end point of progression to AKIN stage 3 and the secondary end points of death and receipt of RRT within 14 days of the FST, determined by assessing the AUC. The AUCs were compared using the method described by DeLong et al.30 and, when appropriate, a Bonferroni adjustment was made to account for multiple comparisons. Additionally, we used logistic regression models consisting of FST and the individual biomarkers and evaluated the discriminatory ability of the combinations to predict patient outcomes compared with the FST alone. Finally, we performed a sequential analysis using biochemical biomarkers and FST, where we first used biomarker cutoffs provided by pre-existing literature for NGAL (>150 ng/ml) and TIMP-2/IGFBP-7 (>0.3). For patients whose biomarker concentration exceeded the cutoff, we then assessed the AUC for FST for all patient outcomes.12,31 We performed all analyses for urinary biomarkers unadjusted (data below) and adjusted for urinary creatinine concentration (data not shown). There was no significant difference between the two.
All means are reported as ±SEM unless otherwise specified. Statistical analysis was performed using SPSS software, version 18.0 (SPSS Inc., Chicago, IL) and Stata 13.0 (StataCorp., College Station, TX). Methods used to calculate FeNa, Acute Physiology and Chronic Health Evaluation II score, and cardiovascular Sequential Organ Failure Assessment score and eGFR were previously described.13
Disclosures
J.L.K. reports consulting fees from Astute Medical Inc. and payments for enrolling patients in observational biomarker studies from Astute and Abbott. A.D.S., reports consulting fees from Astute Medical Inc. L.S.C. reports links to Alere Medical and Astute Medical Inc.
Acknowledgments
The SAKInet portion of this study was funded by National Institutes of Health (NIH), National Institute of Diabetes and Digestive and Kidney Disease (NIDDK), NIH grant number R01-DK080234. J.L.K. was supported by NIDDK NIH grant number K23-DK081616.
Footnotes
Published online ahead of print. Publication date available at www.jasn.org.
See related editorial, “The Furosemide Stress Test and Predicting AKI Outcomes,” on pages 1762–1764.
References
- 1.Chertow GM, Levy EM, Hammermeister KE, Grover F, Daley J: Independent association between acute renal failure and mortality following cardiac surgery. Am J Med 104: 343–348, 1998 [DOI] [PubMed] [Google Scholar]
- 2.Wijeysundera DN, Karkouti K, Dupuis JY, Rao V, Chan CT, Granton JT, Beattie WS: Derivation and validation of a simplified predictive index for renal replacement therapy after cardiac surgery. JAMA 297: 1801–1809, 2007 [DOI] [PubMed] [Google Scholar]
- 3.Koyner JL, Cerdá J, Goldstein SL, Jaber BL, Liu KD, Shea JA, Faubel S, Acute Kidney Injury Advisory Group of the American Society of Nephrology : The daily burden of acute kidney injury: A survey of US nephrologists on World Kidney Day. Am J Kidney Dis 64: 394–401, 2014 [DOI] [PubMed] [Google Scholar]
- 4.Hsu RK, McCulloch CE, Dudley RA, Lo LJ, Hsu CY: Temporal changes in incidence of dialysis-requiring AKI. J Am Soc Nephrol 24: 37–42, 2013 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Faubel S, Chawla LS, Chertow GM, Goldstein SL, Jaber BL, Liu KD, Acute Kidney Injury Advisory Group of the American Society of Nephrology : Ongoing clinical trials in AKI. Clin J Am Soc Nephrol 7: 861–873, 2012 [DOI] [PubMed] [Google Scholar]
- 6.Koyner JL, Parikh CR: Clinical utility of biomarkers of AKI in cardiac surgery and critical illness. Clin J Am Soc Nephrol 8: 1034–1042, 2013 [DOI] [PubMed] [Google Scholar]
- 7.Charlton JR, Portilla D, Okusa MD: A basic science view of acute kidney injury biomarkers. Nephrol Dial Transplant 29: 1301–1311, 2014 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Koyner JL, Vaidya VS, Bennett MR, Ma Q, Worcester E, Akhter SA, Raman J, Jeevanandam V, O’Connor MF, Devarajan P, Bonventre JV, Murray PT: Urinary biomarkers in the clinical prognosis and early detection of acute kidney injury. Clin J Am Soc Nephrol 5: 2154–2165, 2010 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Koyner JL, Garg AX, Coca SG, Sint K, Thiessen-Philbrook H, Patel UD, Shlipak MG, Parikh CR, TRIBE-AKI Consortium : Biomarkers predict progression of acute kidney injury after cardiac surgery. J Am Soc Nephrol 23: 905–914, 2012 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Kashani K, Al-Khafaji A, Ardiles T, Artigas A, Bagshaw SM, Bell M, Bihorac A, Birkhahn R, Cely CM, Chawla LS, Davison DL, Feldkamp T, Forni LG, Gong MN, Gunnerson KJ, Haase M, Hackett J, Honore PM, Hoste EA, Joannes-Boyau O, Joannidis M, Kim P, Koyner JL, Laskowitz DT, Lissauer ME, Marx G, McCullough PA, Mullaney S, Ostermann M, Rimmelé T, Shapiro NI, Shaw AD, Shi J, Sprague AM, Vincent JL, Vinsonneau C, Wagner L, Walker MG, Wilkerson RG, Zacharowski K, Kellum JA: Discovery and validation of cell cycle arrest biomarkers in human acute kidney injury. Crit Care 17: R25, 2013 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Arthur JM, Hill EG, Alge JL, Lewis EC, Neely BA, Janech MG, Tumlin JA, Chawla LS, Shaw AD: Evaluation of 32 urine biomarkers to predict the progression of acute kidney injury after cardiac surgery. Kidney Int 85: 431–438, 2014 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Bihorac A, Chawla LS, Shaw AD, Al-Khafaji A, Davison DL, Demuth GE, Fitzgerald R, Gong MN, Graham DD, Gunnerson K, Heung M, Jortani S, Kleerup E, Koyner JL, Krell K, Letourneau J, Lissauer M, Miner J, Nguyen HB, Ortega LM, Self WH, Sellman R, Shi J, Straseski J, Szalados JE, Wilber ST, Walker MG, Wilson J, Wunderink R, Zimmerman J, Kellum JA: Validation of cell-cycle arrest biomarkers for acute kidney injury using clinical adjudication. Am J Respir Crit Care Med 189: 932–939, 2014 [DOI] [PubMed] [Google Scholar]
- 13.Chawla LS, Davison DL, Brasha-Mitchell E, Koyner JL, Arthur JM, Shaw AD, Tumlin JA, Trevino SA, Kimmel PL, Seneff MG: Development and standardization of a furosemide stress test to predict the severity of acute kidney injury. Crit Care 17: R207, 2013 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Testani JM, Brisco MA, Turner JM, Spatz ES, Bellumkonda L, Parikh CR, Tang WH: Loop diuretic efficiency: A metric of diuretic responsiveness with prognostic importance in acute decompensated heart failure. Circ Heart Fail 7: 261–270, 2014 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Endre ZH, Pickering JW: Biomarkers and creatinine in AKI: The trough of disillusionment or the slope of enlightenment? Kidney Int 84: 644–647, 2013 [DOI] [PubMed] [Google Scholar]
- 16.Parikh CR, Han G: Variation in performance of kidney injury biomarkers due to cause of acute kidney injury. Am J Kidney Dis 62: 1023–1026, 2013 [DOI] [PubMed] [Google Scholar]
- 17.Parikh CR, Coca SG, Thiessen-Philbrook H, Shlipak MG, Koyner JL, Wang Z, Edelstein CL, Devarajan P, Patel UD, Zappitelli M, Krawczeski CD, Passik CS, Swaminathan M, Garg AX, TRIBE-AKI Consortium : Postoperative biomarkers predict acute kidney injury and poor outcomes after adult cardiac surgery. J Am Soc Nephrol 22: 1748–1757, 2011 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Parikh CR, Thiessen-Philbrook H, Garg AX, Kadiyala D, Shlipak MG, Koyner JL, Edelstein CL, Devarajan P, Patel UD, Zappitelli M, Krawczeski CD, Passik CS, Coca SG; for the TRIBE-AKI Consortium: Performance of kidney injury molecule-1 and liver fatty acid-binding protein and combined biomarkers of AKI after cardiac surgery. Clin J Am Soc Nephrol 8: 1079–1088, 2013 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Palevsky PM, Molitoris BA, Okusa MD, Levin A, Waikar SS, Wald R, Chertow GM, Murray PT, Parikh CR, Shaw AD, Go AS, Faubel SG, Kellum JA, Chinchilli VM, Liu KD, Cheung AK, Weisbord SD, Chawla LS, Kaufman JS, Devarajan P, Toto RM, Hsu CY, Greene T, Mehta RL, Stokes JB, Thompson AM, Thompson BT, Westenfelder CS, Tumlin JA, Warnock DG, Shah SV, Xie Y, Duggan EG, Kimmel PL, Star RA: Design of clinical trials in acute kidney injury: Report from an NIDDK workshop on trial methodology. Clin J Am Soc Nephrol 7: 844–850, 2012 [DOI] [PubMed] [Google Scholar]
- 20.Okusa MD, Molitoris BA, Palevsky PM, Chinchilli VM, Liu KD, Cheung AK, Weisbord SD, Faubel S, Kellum JA, Wald R, Chertow GM, Levin A, Waikar SS, Murray PT, Parikh CR, Shaw AD, Go AS, Chawla LS, Kaufman JS, Devarajan P, Toto RM, Hsu CY, Greene TH, Mehta RL, Stokes JB, Thompson AM, Thompson BT, Westenfelder CS, Tumlin JA, Warnock DG, Shah SV, Xie Y, Duggan EG, Kimmel PL, Star RA: Design of clinical trials in acute kidney injury: Areport from an NIDDK workshop—prevention trials. Clin J Am Soc Nephrol 7: 851–855, 2012 [DOI] [PubMed] [Google Scholar]
- 21.Molitoris BA, Okusa MD, Palevsky PM, Chawla LS, Kaufman JS, Devarajan P, Toto RM, Hsu CY, Greene TH, Faubel SG, Kellum JA, Wald R, Chertow GM, Levin A, Waikar SS, Murray PT, Parikh CR, Shaw AD, Go AS, Chinchilli VM, Liu KD, Cheung AK, Weisbord SD, Mehta RL, Stokes JB, Thompson AM, Thompson BT, Westenfelder CS, Tumlin JA, Warnock DG, Shah SV, Xie Y, Duggan EG, Kimmel PL, Star RA: Design of clinical trials in AKI: A report from an NIDDK workshop. Trials of patients with sepsis and in selected hospital settings. Clin J Am Soc Nephrol 7: 856–860, 2012 [DOI] [PubMed] [Google Scholar]
- 22.Bonventre JV, Basile D, Liu KD, McKay D, Molitoris BA, Nath KA, Nickolas TL, Okusa MD, Palevsky PM, Schnellmann R, Rys-Sikora K, Kimmel PL, Star RA, Kidney Research National Dialogue (KRND) : AKI: A path forward. Clin J Am Soc Nephrol 8: 1606–1608, 2013 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Alge JL, Karakala N, Neely BA, Janech MG, Tumlin JA, Chawla LS, Shaw AD, Arthur JM, Investigators SA, SAKInet Investigators : Association of elevated urinary concentration of renin-angiotensin system components and severe AKI. Clin J Am Soc Nephrol 8: 2043–2052, 2013 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Alge JL, Karakala N, Neely BA, Janech MG, Tumlin JA, Chawla LS, Shaw AD, Arthur JM, Investigators SA, SAKInet Investigators : Urinary angiotensinogen and risk of severe AKI. Clin J Am Soc Nephrol 8: 184–193, 2013 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Mehta RL, Kellum JA, Shah SV, Molitoris BA, Ronco C, Warnock DG, Levin A, Acute Kidney Injury Network : Acute Kidney Injury Network: Report of an initiative to improve outcomes in acute kidney injury. Crit Care 11: R31, 2007 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Chawla LS, Dommu A, Berger A, Shih S, Patel SS: Urinary sediment cast scoring index for acute kidney injury: A pilot study. Nephron Clin Pract 110: c145–c150, 2008 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Parikh CR, Devarajan P, Zappitelli M, Sint K, Thiessen-Philbrook H, Li S, Kim RW, Koyner JL, Coca SG, Edelstein CL, Shlipak MG, Garg AX, Krawczeski CD, TRIBE-AKI Consortium : Postoperative biomarkers predict acute kidney injury and poor outcomes after pediatric cardiac surgery. J Am Soc Nephrol 22: 1737–1747, 2011 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Chaturvedi S, Farmer T, Kapke GF: Assay validation for KIM-1: Human urinary renal dysfunction biomarker. Int J Biol Sci 5: 128–134, 2009 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Coca SG, Jammalamadaka D, Sint K, Philbrook HT, Shlipak MG, Zappitelli M, Devarajan P, Hashim S, Garg AX, Parikh CR: Preoperative proteinuria predicts acute kidney injury in patients undergoing cardiac surgery. J Thorac Cardiovasc Surg 143: 495–502, 2012 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.DeLong ER, DeLong DM, Clarke-Pearson DL: Comparing the areas under two or more correlated receiver operating characteristic curves: A nonparametric approach. Biometrics 44: 837–845, 1988 [PubMed] [Google Scholar]
- 31.de Geus HR, Bakker J, Lesaffre EM, le Noble JL: Neutrophil gelatinase-associated lipocalin at ICU admission predicts for acute kidney injury in adult patients. Am J Respir Crit Care Med 183: 907–914, 2011 [DOI] [PubMed] [Google Scholar]