Utilizing reclassification to explore characteristics and prognosis of KDIGOSCr AKI subgroups: a retrospective analysis of a multicenter prospective cohort study

Abstract Background Acute kidney injury (AKI) is widespread in the intensive care unit (ICU) and affects patient prognosis. According to Kidney Disease: Improving Global Outcomes (KDIGO) guidelines, the absolute and relative increases of serum creatinine (Scr) are classified into the same stage. Whether the prognosis of the two types of patients is similar in the ICU remains unclear. Methods According to the absolute and relative increase of Scr, AKI stage 1 and stage 3 patients were divided into stage 1a and 1b, stage 3a and 3b groups, respectively. Their demographics, laboratory results, clinical characteristics, and outcomes were analyzed retrospectively. Results Of the 345 eligible cases, we analyzed stage 1 because stage 3a group had only one patient. Using 53 or 61.88 µmol/L as the reference Scr (Scrref), no significant differences were observed in ICU mortality (P53=0.076, P61.88=0.070) or renal replacement therapy (RRT) ratio, (P53=0.356, P61.88=0.471) between stage 1a and 1b, but stage 1b had longer ICU length of stay (LOS) than stage 1a (P53<0.001, P61.88=0.032). In the Kaplan-Meier survival analysis, no differences were observed in ICU mortality between stage 1a and 1b (P53=0.378, P61.88=0.255). In a multivariate analysis, respiratory failure [HR = 4.462 (95% CI 1.144–17.401), p = 0.031] and vasoactive drug therapy [HR = 4.023 (95% CI 1.584–10.216), p = 0.003] were found to be independently associated with increased risk of death. Conclusion ICU LOS benefit was more prominent in KDIGOSCr AKI stage 1a patients than in stage 1 b. Further prospective studies with a larger sample size are necessary to confirm the effectiveness of reclassification.


Introduction
Acute kidney injury (AKI) comprises a heterogeneous group of conditions characterized by a sudden decrease in the glomerular filtration rate, manifested by an increase in serum creatinine concentration or oliguria, and classified by stage and cause [1]. AKI itself might independently increase mortality, and it is associated with other negative consequences, such as progression to chronic kidney disease, which may require renal replacement therapy (RRT), prolonged hospitalization, increased medical costs, and subsequent lower quality of life [2][3][4][5][6][7]. Studies have reported that in the intensive care unit (ICU) setting, AKI-associated morbidity and mortality rates are 55.38-57.3% and 25.8-26.9%, respectively [8][9]. Because serum creatinine (Scr) level is highly associated with the outcome in patients with AKI [10], international consensus criteria have been developed and later refined for the diagnosis and staging of AKI, the severity of which is classified according to urine output and elevations in Scr level [11][12][13]. The recent Kidney Disease: Improving Global Outcomes (KDIGO) clinical practice guidelines defined AKI stage 1 as the following: increase in Scr by !26.5 mmol/l within 48 h; or an increase in Scr to !1.5 times of baseline, which is known or presumed to have occurred within the preceding 7 days [11]. However, whether the prognosis of the two types of patients in AKI stage 1 is consistent in ICU remains unclear.
Recently, Sparrow et al. [14] evaluated the potential impact of further categorizing AKI stage 1 into two stages based on Scr criteria in a cohort of 81,651 inpatients, that is, AKI stage 1a as an absolute increase in Scr of 26.5 mmol/L (0.3 mg/dl) within 48 h and stage 1b as a 50% relative increase in Scr within 7 days. The authors found that patients with AKI stages 1a and 1b experienced clinically meaningful and statistically significant differences in length of stay (LOS) and mortality. This study suggests that a modified 2-stage version of the KDIGO AKI stage 1 may provide additional prognostic information.
At present, there is no detailed research on characteristics of hospitalized patients in China based on further categorizing AKI stages, especially ICU patients. Therefore, we aimed to investigate the influence of such a strategy of further categorizing AKI stage 1 on the clinical prognosis of AKI patients in the ICU setting based on the Chinese critical care trial group database.

Study design and setting
The prospective observational study was performed from 1 July 2009 to 31 August 2009 in 22 tertiary hospitals from 19 provinces and autonomous regions of China [15]. We conducted a retrospective study with information from this prospective cohort database. All patients who were admitted to the ICU followed the guidelines for the construction and management of critical care departments (Trial) issued by the Ministry of health of China in 2009 [16]. Initial database research was approved by the ethics committee of the Fuxing Hospital affiliated to Capital Medical University. The data, including patient records and information, were anonymized and de-identified prior to analysis. Written informed consent was waived because of the study design. The STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guideline recommendations were used as a reference [17].

Standards and definitions
Baseline Scr (Scr baseline ) refers to the Scr measured on hospital admission, while Scr ref refers to the lowest Scr value within 48 h or 7 days, and peak Scr (Scr peak ) refers to the highest Scr value when AKI is diagnosed.
The proposed modifications to KDIGO AKI Scr (KDIGO Scr -AKI) stage are shown in Table 1. According to the absolute or relative increase of Scr, stage 1 was divided into two subgroups: stage 1a and stage 1b. Stage 3 patients also followed this method and were divided into two subgroups: stage 3a and stage 3b. The Scr absolute increase of 26.5 mmol/L is equal to a 50% increase when Scr ref is 53 mmol/L. According to the KDIGO Scr -AKI staging standard [11], stages can be rewritten as the formula: y ¼ kx þ e, stage 1a: y ! x þ 26.5 and y < 1.5x; stage 1 b: y ¼ ax, 1.5 a < 2; stage 2: y ¼ bx, 2 b < 3; stage 3a: y ¼ xe þ 44.2, x ! 353.6; stage 3 b: y ¼ cx, c ! 3. If x þ 26.5 ¼ 1.5x, then  If a patient met stage 1a criteria on hospital day 2 and progressed to stage 1 b within 7 days, the patient was classified as stage 1 b. In order to further illustrate whether the standard of our hypothesis is correct or not, we selected Scr ref ¼ 61.88 mmol/L [14] to reanalyze our data.

Clinical variables
Demographic and clinical data were collected at ICU presentation, including gender, age, weight, acute physiology and chronic health evaluation II (APACHE II) score, sequential organ failure assessment (SOFA) score, Charlson comorbidity index (CCI) [18], estimated glomerular filtration rate (eGFR) [19], admission status, reasons for ICU admission, interventions during ICU, ICU LOS and comorbidities. Scr levels were recorded for one week after admission.
Statistical analysis SPSS 25.0 (SPSS Inc., Chicago, IL) and OriginPro8 mapping software (OriginLab Inc. Northampton, MA) were used for data analysis. Kolmogorov-Smirnov test was used to test for normality. Continuous variables were described as the median (interquartile range, IQR) and were compared by the Mann-Whitney U test. Categorical variables were presented as n (%) and were compared by the chi-square test or Fisher's exact test. We assessed the risk of patients' death based on refined stage 1 and using the Kaplan-Meier curves and the log-rank test, followed by multivariable-adjusted Cox proportional hazards models adjusted for covariates and potential confounders. All statistical analyses were performed using the bicaudal test, p < 0.05 was considered statistically significant.

General information
The flow diagram of the study is shown in Figure 2. From a total of 3063 records from database, reasons for exclusion included age 18 years (n ¼ 127), ICU LOS <24 h (n ¼ 1623), no Scr baseline (n ¼ 18), end-stage kidney disease (n ¼ 59), kidney transplantation (n ¼ 1), fewer than 2 Scr measurements in ICU (n ¼ 24), without progression to AKI (n ¼ 697), Scr ref <53 mmol/L (n ¼ 128) and incomplete clinical data (n ¼ 41). Based on the existing database, the available sample size is 345 who were all enrolled in the final analysis.

Demographics and characteristics of patients
The demographics and general characteristics of the 345 patients included in the study are reported in Table  2. The median age of the patients was 59 years (IQR 42-73), and 78.13% of the participants were male. The    (Figure 4). Figure 5 demonstrates using the univariate Kaplan-Meier survival analysis, patient survival was not found to be statistically different between stage 1a and stage 1b patients (log-rank Test:

Survival analyses
This result remained non-significant after re-classification with Scr ref ¼ 61.88 mmol/L (log-rank Test: We chose the multivariate Cox regression hazard model to test for differences in the hazard of death over 28 days according to refined stage 1, in order to allow for the correction of potential confounding factors including age, eGFR, APACHE II, SOFA, chronic kidney disease, CCI, sepsis, trauma, surgery and mechanical ventilation, respiratory failure and vasoactive drug therapy as covariates, which were found to be independently associated with ICU mortality from univariate Kaplan-Meier survival analysis. Respiratory   Other variables included in the analysis were not found to be independently associated with ICU mortality (Table 3).

Discussion
AKI is defined by a rapid increase in serum creatinine, decreased urine output, or both [20]. Since the KDIGO guideline for AKI was published in 2012 [11], substantial advances in our understanding of AKI epidemiology, pathophysiology, and diagnostic testing have fueled a growing controversy. However, the concept of AKI staging has clear and significant limitations that should be addressed, as it has relied on the established but poor biomarkers of solute clearance (serum creatinine levels and urinary output), and has been challenged by the identification of novel biomarkers of tubular stress and damage. However, the AKI criteria continue to be valuable, when no acceptable alternative was available [21].
The present study demonstrates that while stage 1b has the better basic renal function (higher eGFR), we found that the two subgroups differed significantly only in ICU LOS, however the two Scr ref criteria (53 mmol/L or 61.88 mmol/L) in KDIGO AKI stage 1 did not distinguish the two associated populations in ICU mortality or RRT support. Furthermore, we could not establish an independent association of reclassification of stage 1 to ICU mortality. Respiratory failure and vasoactive drug therapy were found to be independently associated with the increased risk for death. Our results differed from that of a recent study [14], in which Sparrow and his colleagues screened 81,651 patients admitted to a large academic medical center and 4 satellite community hospitals. To operationalize the proposed 4-stage criteria correctly, they used linear regression and determined that the lower bound for Scr ref was 61.88 mmol/L, and the LOS for stage 1b was longer than stage 1a. Moreover, in-hospital mortality was found to increase as the severity of AKI increased. Patients with AKI stages 1a and 1b experienced clinically significant differences in the LOS and mortality.
Our results showed no differences in mortality based on refined staging KDIGO Scr 22.61% of AKI patients and the statistical difference between Scr peak 1b and Scr peak 1a is significant. Second, the characteristics of ICU patients are different from those in the general ward. ICU patients may be admitted because of acute respiratory distress syndrome, septic shock, multiple trauma and many other different reasons, and there is often multiple organ damage. Although studies have found that a 1.5-fold increase in Scr during steady state conditions reflects a 39% decrease in eGFR [23], and the mortality rate increased to 6% in patients whose Scr levels increased to 44.2 mmol/L [24], but the pooled mortality rises to 42% in critically ill patients with KDIGO stage 3 and 46% of those requiring RRT [25]. KDIGO AKI stage 1, as the early impairment of a single organ can predict functional changes but which might not quantify damage, is unlikely to affect mortality or RRT ratio. Mortality in the Sparrow's study was 49.7%, while in our study was 13.91%. This reduction in the total number of deaths in each stage, make it more difficult to find statistical differences in the current study.
There are some limitations to our study. First, this was a retrospective analysis of prospectively collected data, the database has been established for a long time, and the sample size was limited so we failed to analyze stage 3; unknown confounders could have affected the study results. The results of our study should be interpreted cautiously and require application in much larger ICU populations to elucidate further whether significant differences exist.
To conclude, we explored the characteristics and prognosis of KDIGO SCr AKI stage 1. We found that stage 1a patients was beneficial in terms of ICU LOS compared to stage 1b when Scr ref is 53 mmol/L or higher, but stage 1a patients had not decreased ICU mortality and RRT support.

Disclosure statement
No potential conflict of interest was reported by the author(s).

Data availability statement
The datasets and resources analyzed during the current study are available from the corresponding author upon reasonable request.