Utility of Biomarkers for Sepsis-Associated Acute Kidney Injury Staging

This cohort study examines associations between urinary levels of 2 cell-cycle arrest biomarkers and 30-day survival among patients with septic shock who developed acute kidney injury staged based on kidney function.


Introduction
Acute kidney injury (AKI) is defined by an abrupt decrease of the glomerular filtration rate and has multiple causes.Surrogates for the glomerular filtration rate, specifically serum creatinine level and urine output, are used to stage AKI severity according to Kidney Disease: Improving Global Outcomes (KDIGO) criteria. 1Although there is good correlation between AKI stage and histologic findings on kidney biopsy, 2 there can be kidney damage without change in function 3 and injury not detected by AKI criteria.Furthermore, as in chronic kidney disease, the presence or absence of damage markers might add to severity grading in AKI when functional changes are also present. 4][7][8][9][10] However, their evaluation and commercial development have been focused on predicting functional change, not on assessing damage or augmenting AKI staging.The only US Food and Drug Administration-approved test for AKI, NephroCheck Test (Astute Medical, Inc), measures 2 urinary biomarkers, tissue inhibitor of metalloproteinases 2 (TIMP-2) and insulinlike growth factor binding protein 7 (IGFBP7), combining the product of the concentration of the 2 biomarkers in the score [TIMP-2] × [IGFBP7]. 11These molecules are released from kidney tubular cells even with sublethal noxious stimuli, such as a nutrient deprivation in vitro 12 or remote ischemic preconditioning in vivo. 13Whether such stress markers identify damage at higher thresholds is uncertain.
In 2020, the 23rd Acute Disease Quality Initiative (ADQI-23) work group proposed an expanded classification for AKI, adding biomarkers for each stage (Table 1). 10,146][17] Moreover, compared with biomarker-negative patients without standard functional criteria for AKI, higher levels of [TIMP-2] × [IGFBP7] are associated with a reduction in kidney functional reserve. 18The ADQI-23 work group specified the criteria for qualifying biomarkers to aid in the classification of AKI (rather than to detect it). 10 In brief, a biomarker candidate must differentiate patients with a different prognosis at a given functional stage.In this cohort study of critically ill patients with septic shock from the Protocolized Care for Early Septic Shock (ProCESS) trial, 19 we explored whether higher levels of [TIMP-2] × [IGFBP7], measured nearly simultaneously with assessment of AKI, could identify patients with reduced survival among those at the same functional AKI stage.

Study Design and Patients
This cohort study used existing data from the ProCESS trial.The ProCESS trial design, methods, and main results have been described elsewhere. 19,20 We collected patient demographics, health history, and serum creatinine level (12 months before and at enrollment and then daily until discharge from the hospital or truncated at 28 days) and urine output (hourly for the first 72 hours or until discharge from the intensive care unit) values as previously reported. 19We classified AKI according to KDIGO criteria using both serum creatinine level and urine output (eAppendix 2 in Supplement 1). 1 We obtained urine samples for biomarker testing at 6 hours after enrollment and the start of resuscitation.We centrifuged the samples immediately after collection and froze and stored the supernatant at less than -70 °C.We thawed the supernatant immediately before testing for the [TIMP-2] × [IGFBP7] level with the NephroCheck Test.Tests were performed according to the manufacturer's specifications.

Exposures and End Points
We evaluated the presence of AKI based on KDIGO criteria between 0 and 24 hours after enrollment and the highest AKI stage and the urinary [TIMP-2] × [IGFBP7] level at 6 hours after enrollment.To augment AKI diagnosis and staging, we applied a previously reported high-specificity cutoff for [TIMP-2] × [IGFBP7] of 2.0 (ng/mL) 2 /1000 15,22 to categorize patients according to the new AKI staging system proposed by ADQI-23 (Table 1). 10r primary end point was survival 30 days after enrollment.Our secondary end points were (1)   all-cause mortality 30 days after enrollment; (2) full recovery from AKI (for patients who developed AKI at any stage in the first 24 hours), defined as the absence of any AKI on the last day available (truncated at 28 days after enrollment); (3) AKI stage 3 by day 7 (for patients who did not develop AKI stage 3 in the first 24 hours), defined as reaching AKI stage 3 in any day between day 1 (excluded) and day 7 (included) after enrollment; and (4) hospital length of stay (LOS), truncated at 60 days after enrollment among patients who were alive at 60 days.

Statistical Analysis
We compared survival 30 days after enrollment among patients at the new AKI stages obtained with the [TIMP-2] × [IGFBP7] level using Kaplan-Meier plots and the log-rank test.We also used Cox proportional hazards regression models to obtain adjusted hazard ratios (HRs) for death within 30 days and corresponding adjusted survival curves.To adjust survival curves, we used covariates present before the occurrence of the exposures, such as age, sex, race and ethnicity, and Charlson Comorbidity Index (to assess the burden of previous comorbidities).
For the secondary end points, we expressed categorical variables as absolute numbers and percentages and compared them using Pearson χ Lehmann median difference (95% CI) as an effect measure.We did not apply corrections for multiple comparisons regarding the 3 secondary end points because they were considered as exploratory.We conducted all analyses using SPSS, version 26 (IBM Corp), and we set the per-comparison significance at 2-sided P < .05.
We performed a sensitivity analysis evaluating associations between the end points and cutoffs of 0.3 and 1.0 (ng/mL) 2 /1000 for the [TIMP-2] × [IGFBP7] level.The 0.3 (ng/mL) 2 /1000 cutoff is considered a high-sensitivity cutoff based on the results of both its derivation 22 and validation studies, 23 whereas the 1.0 (ng/mL) 2 /1000 cutoff has been evaluated for sepsis and has shown good sensitivity and specificity compared with the 0.3 and 2.0 (ng/mL) 2 /1000 cutoffs 24 (eAppendix 3 in Supplement 1).

Results
The analysis cohort included 999 patients for whom the

Discussion
Many investigators have suggested implementing biomarkers in the evaluation of AKI, 8,9,25 and the ADQI-23 consensus conference 10 provided a conceptual framework for expanding functional classification of AKI by adding biomarkers.The cell-cycle arrest biomarkers TIMP-2 and IGFBP7 are produced by kidney tubule epithelial cells in response to kidney stress. 5,9A [TIMP-2] × [IGFBP7] level greater than 2.0 (ng/mL) 2 /1000 has high specificity (95%) for predicting moderate to severe AKI 23 ; thus, we used this cutoff for our primary analysis.Lower cutoffs of 0.3 and 1.0 (ng/mL) 2 /1000 have been reported in the literature [22][23][24] ; they have higher sensitivity for detecting AKI, but for staging a disease or syndrome, high specificity is needed.For this reason, we chose 2.0 (ng/mL) 2 /1000 as the cutoff for our primary analysis.
The ADQI-23 consensus recommendations emphasized the need for biomarkers to add prognostic information to existing function-based AKI staging. 10However, it is important that this information be specific to AKI rather than an overall prognostic marker.In this study, a [TIMP-2] × [IGFBP7] level greater than 2.0 (ng/mL) 2 /1000 measured nearly simultaneously with the assessment of AKI was associated with lower survival among patients at all 3 AKI stages but not among patients without AKI (stage 1S), although patients in this group had a longer median hospital LOS if their [TIMP-2] × [IGFBP7] level was greater than 2.0 (ng/mL) 2 /1000.Similar to these findings, prior studies showed that the rate of death or dialysis by 9 months 15,17 or by hospital discharge 16 was increased in patients positive for a biomarker only when functional changes also occurred.As such, the prognostic signal would seem to be specific to AKI and would not be a general mortality indicator.
The results are also consistent with the nature of TIMP-2 and IGFBP7 as stress rather than damage biomarkers.Stress in isolation may be benign, but when it coexists with dysfunction, it may signal actual damage. 17However, this interpretation is challenged by the work of Husain-Syed and colleagues, 18 who demonstrated that an increased urinary [TIMP-2] × [IGFBP7] level after cardiac surgery was associated with loss of kidney functional reserve at 3 months even in the absence of AKI.
Thus, crude measures, such as death or dialysis, may not help detect subclinical injury, but measuring kidney functional reserve may help reveal it.It is also possible that in certain conditions (eg, loss of muscle mass), creatinine levels may fail to accurately reflect kidney function.With regard to our secondary end points, we also found that for certain functional stages (no AKI and stage 2), a [TIMP-2] × [IGFBP7] level greater than 2.0 (ng/mL) 2 /1000 was associated with a longer hospital LOS and, only for patients with stage 2, a lower rate of full recovery from AKI.However, caution must be used when interpreting the results of the secondary end points because they were exploratory and should only be used to guide future studies using more kidney-related end points.
To our knowledge, this is the first study to specifically examine the conceptual framework recently put forth by the ADQI-23 work group.However, we are not the first to examine how biomarkers might add information to standard functional stages.Uhel and colleagues 26 examined 16 plasma biomarkers reflecting pathways involved in sepsis pathogenesis and blood leukocyte transcriptomes in patients with sepsis.They found associations between these biomarker signatures and both AKI occurrence and persistence, as defined by ADQI-16, 27 finding significant differences in both.We chose to focus on the only commercially available biomarker test for AKI in the US because it is also available worldwide and has a published high-specificity cutoff and because an ongoing trial is already using it specifically for patients with sepsis. 28Future work is needed to evaluate other potential biomarkers alone or in combination to augment AKI staging, such as kidney injury molecule 1 and neutrophil gelatinase-associated lipocalin.Prior evidence suggests that kidney injury molecule 1 may be able to detect kidney damage below the serum creatinine threshold 7,29,30 and that patients positive for neutrophil gelatinase-associated lipocalin, even without the presence of serum creatinine criteria for AKI, were at greater risk of death and/or kidney replacement therapy.or 12 hours 35 after initial therapy in patients with sepsis.Fourth, because of the exploratory and retrospective nature of our analysis, we could not determine sample sizes a priori and the study was likely underpowered for certain substages; we also could not fully adjust for multiple comparisons (multiple secondary end points and cutoffs).Our secondary end points included variables that were not adjusted for competing risk of death.Fifth, we did not have data on long-term kidney function (to calculate major adverse kidney events) and instead used 30-day survival as our main end point for evaluating the biomarker.Although we believe this end point is appropriate for septic shock, it would not be appropriate for conditions with low mortality (eg, after cardiac surgery).The use of more kidney-related secondary end points was motivated by an attempt to reduce this limitation.

Conclusions
In this cohort study, integration of biomarkers of cell-cycle arrest in the staging process of AKI as

Figure 2 .
Figure 2. Covariate-Adjusted Survival Among Patients at Each Functional Acute Kidney Injury (AKI) Stage

Table 1 .
Proposed New Definition and Staging of Acute Kidney Injury by the ADQI-23 Consensus Conference a 14 conversion factor: To convert serum creatinine to micromoles per liter, multiply by 76.25.aAdapted from the ADQI-2314and published by Ostermann et al. 10

JAMA Network Open | Critical Care Medicine
Utility of Biomarkers for Sepsis-Associated Acute Kidney Injury Staging 21e ProCESS trial was a multicenter, randomized clinical trial of 3 different resuscitation strategies in patients with septic shock that enrolled 1341 patients at academic and community emergency departments and intensive care units in the US from March 2008 to May 2013.We excluded patients with end-stage kidney disease (receiving long-This retrospective study was performed in accordance with the 1964 Declaration of Helsinki.21TheOffice of Human Research Protection at the University of Pittsburgh approved this ancillary analysis, which was conducted using deidentified data.All patients or their legally authorized representatives provided written informed consent.This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.Data were analyzed from October 2020 to October 2021.

Downloaded From: https://jamanetwork.com/ on 09/20/2023 compared
them using the Mann-Whitney U test together with the independent-samples Hodges-

Table 1
. Covariates used in the Cox proportional hazards regression models were age, sex, race and ethnicity, and Charlson Comorbidity Index.Dashed lines indicate patients who were biomarker negative, and solid lines indicate those who were biomarker positive.

Table 3
summarizes the secondary end points according to each KDIGO stage and compares them

Table 3 .
Secondary End Points Compared Between Participants Within the Same Functional KDIGO AKI Stage Who Were Negative vs Positive for Biomarkers and eTable 3 in Supplement 1), whereas the 0.3 (ng/mL) 2 /1000 cutoff differentiated survival only between stages 2A and 2B (eFigures 4 and 5 and eTable 4 in Supplement 1).