Predictive Value of Anion Gap for Acute Kidney Injury Progression and Outcomes

Introduction

The AG is an index calculated from serum electrolyte measurement values, used to assess the cause of metabolic acidosis [1]. It represents the difference in charge between the major cations and major anions in the blood [2]. Metabolic acidosis is classified into 2 categories based on the presence of unmeasured anions in serum: high AG metabolic acidosis and normal AG metabolic acidosis [3]. Elevated AG indicates greater metabolic derangements, tissue hypoperfusion, and systemic inflammation, which are associated with the severity of AKI, multi-organ damage, and increased mortality [4]. AKI is a critical health problem with significant morbidity and mortality across various clinical settings, affecting millions globally [5]. It is defined by an abrupt decline in kidney function evidenced by changes in markers such as serum creatinine and urine output [6]. Even with modern dialysis techniques, AKI continues to be associated with major adverse outcomes, including chronic kidney disease progression, end-stage renal disease, and premature death [7]. The severity of AKI ranges from minor, potentially self-resolving injuries to severe cases necessitating dialysis and significantly heightening the risk of mortality [8]. Thus, early detection and continuous monitoring of AKI progression and outcome are critical. Recently, there has been an increased focus on the relationship between the AG and AKI. Most studies have concentrated on using various factors to anticipate AKI risk in critically ill patients [7,9–12], and less research has been conducted on the progression and prognosis of AKI, particularly in non-critically ill patients. Moreover, most prior studies used initial AG measurements, limiting the relevance of their conclusions to the prognosis of patients with AKI. It is therefore imperative to explore how AG at the onset of AKI correlates with its subsequent progression and patient prognosis.

Material and Methods

OBJECTIVE:

This study explored the relationship between the AG at the onset of AKI and the progression and prognosis of AKI.

STUDY DESIGN AND DATA SOURCE:

The research was based on a retrospective analysis of data from a multicenter randomized trial on an AKI early warning system. The data was sourced from an online repository [13], Dryad (https://datadryad.org/stash/dataset/doi:10.5061/dryad.4f4qrfj95) [29].

ETHICS APPROVAL AND CONSENT TO PARTICIPATE:

The investigations involving human participants received approval from the Research Ethics Committee of the Yale School of Medicine institutional review board (ID 1604017596) and the Bridgeport Hospital institutional review board (IDs 051802, 041801, 071808) under a waiver of informed consent. The study was performed in accordance with the amended Declaration of Helsinki and our study involved retrospective data analysis.

SETTING:

The study involved adult inpatients from 6 hospitals in the Yale New Haven Health System.

INCLUSION AND EXCLUSION CRITERIA:

The inclusion criteria were patients aged 18 years and above who were diagnosed with AKI based on the KDIGO (Kidney Disease: Improving Global Outcomes) AKI criteria. We excluded patients with previous dialysis, end-stage renal disease, baseline serum creatinine ≥4.0 mg/L, hospice care, kidney transplantation within the last 6 months, missing and extreme values of the AG at the onset of AKI (greater than or less than 3 standard deviations), and AG at the onset of AKI measured beyond more than 6 hours after the identification of AKI.

PARTICIPANTS:

The study enrolled 5731 individuals with AKI from March 29, 2018, to December 14, 2019, who met the inclusion and exclusion criteria.

OUTCOME INDICATORS:

The outcomes of the study included 14-day AKI progression, 14-day dialysis, and14-day mortality. We defined 14-day AKI progression as a rise in the AKI stage within 14 days based on the KDIGO AKI criteria within 14 days.

STATISTICAL ANALYSIS:

Continuous variables were summarized by calculating the median and range (minimum to maximum), while categorical variables were represented using counts and percentages. Univariate analysis, multivariate logistic regression, and smooth curve fitting (generalized additive modeling) were used to evaluate the association between AG at the onset of AKI and AKI progression and prognosis of patients with AKI. Since the Elixhauser score included comorbidities and the SOFA score incorporated platelets and blood creatinine, variables related to these factors were excluded from the multivariate analysis to avoid collinearity after incorporating the Elixhauser and SOFA scores. A two-sided P<0.05 significance level was used. Statistical analysis was carried out using EmpowerStats software.

Results

THE CLINICAL CHARACTERISTIC OF PATIENTS:

All laboratory tests, such as creatinine and blood urea nitrogen, were conducted at the onset of AKI, within 6 hours of its diagnosis. A total of 6030 patients were initially included in the primary study. Missing and extreme values (greater than or less than 3 standard deviations) of the AG at the onset of AKI were excluded, removing 132 patients. Additionally, measuring AG at the onset of AKI more than 6 hours after AKI identification excluded 167 more individuals. Finally, a total 5731 patients were included in our study. The median age was 69.45 years (range: 18.14–89.99 years). There were slightly more male (52.10%) than female (47.90%) patients. Most patients were of the “other” race (84.33%), with 15.67% being Black. Most were admitted to urban teaching hospitals (78.05%), followed by suburban non-teaching (12.48%) and suburban teaching (9.47%) hospitals. Medical admissions accounted for 75.43% patients, 50.64% were on alert status, with the rest (49.36%) receiving usual care. Over 30% of them were admitted to the ICU. Comorbid conditions were common, including chronic kidney disease (38.21%), COPD (34.43%), diabetes (41.39%), hypertension (82.38%), and congestive heart failure (44.72%), and 19.96% had recent contrast agent exposure. The median bicarbonate level was 24.00 mmol/L, with a wide range from 7.00 to 50.00 mmol/L. Blood urea nitrogen (BUN) levels had a median of 28.00 mg/dL, ranging from 1.00 to 182.00 mg/dL. Hemoglobin levels were reported with a median of 10.60 g/dL, spanning from 4.00 to 21.70 g/dL. Estimated glomerular filtration rate (eGFR) had a median of 55.76 mL/min, with a range indicating a broad spectrum of kidney function across the cohort (9.59–294.66 mL/min). Potassium (K+) levels had a median of 4.20 mmol/L, and sodium (Na+) levels had a median of 138.00 mmol/L. White blood cell (WBC) counts varied significantly, with a median of 9.80×1000/μL. AG at the onset of AKI had a median of 12 mmol/L, ranging from 1.00 mmol/L to 25 mmol/L. The median Elixhauser and SOFA scores were 6 and 2, respectively. In the 14 days following admission, 23.63% had a nephrology consultation. Most AKI episodes (71.98%) lasted <2 days. The 14-day AKI progression occurred in 15.51% of the patients, while 3.07% required dialysis, and 8.32% died (Figure 1, Table 1).

RESULTS OF MULTIVARIATE LOGISTIC REGRESSION ANALYSIS, EXAMINING THE ASSOCIATION BETWEEN VARIOUS EXPOSURES AND AKI PROGRESSION AND PROGNOSIS OF PATIENTS WITH AKI:

In the multivariate logistic regression analysis (Table 2), we examined the associations between various clinical exposures and the progression and prognosis of acute kidney injury (AKI) within a 14-day period. The results revealed several significant predictors. An elevated AG at the onset of AKI was consistently associated with worse outcomes across all measured endpoints. For every unit increase in the AG at the onset of AKI (mmol/L), there was a 7% increase in the odds of AKI progression (OR 1.07, 95% CI 1.04 to 1.10, P<0.01), a 9% increase in the likelihood of requiring dialysis (OR 1.09, 95% CI 1.04 to 1.15, P<0.01), and an 11% increase in mortality risk (OR 1.11, 95% CI 1.07 to 1.14, P<0.01). Age was significantly associated with 14-day mortality, but not with 14-day AKI progression or 14-day dialysis. Sex and race were not significantly associated with any of the outcomes. Admit medical status was associated with higher 14-day mortality. Patients admitted to suburban teaching and suburban non-teaching hospitals had a significantly higher risk of 14-day mortality compared to those admitted to urban teaching hospitals. Alert status and hemoglobin levels were not significantly associated with any of the outcomes. BUN level was associated with 14-day AKI progression and 14-day mortality. An increase in the eGFR was significantly associated with a higher risk of 14-day AKI progression and 14-day mortality, but was associated with a lower risk of 14-day dialysis. ICU admission was significantly associated with a higher risk of 14-day AKI progression and 14-day mortality. Patients who had a nephrology consult within 14 days had a significantly higher risk of all outcomes. Contrast agent use in the prior 72 hours was significantly associated with a higher risk of 14-day mortality. Patients with AKI duration of 2 to 7 days and more than 7 days had a significantly higher risk of 14-day AKI progression compared to those with AKI duration of less than 2 days. The Elixhauser score and SOFA score were significantly associated with a higher risk of all outcomes, except for the Elixhauser score’s association with 14-day mortality (Table 2).

MULTIVARIATE LOGISTIC REGRESSION ANALYSIS OF THE RELATIONSHIP BETWEEN THE AG AT THE ONSET OF AKI AND THE AKI PROGRESSION AND THE PROGNOSIS OF PATIENTS WITH AKI:

The multivariate logistic regression analysis revealed a significant association between AG at the onset of AKI and progression and prognosis of acute kidney injury (AKI) in patients. In the crude model, an increase in the AG at the onset of AKI was significantly associated with a higher risk of 14-day AKI progression (OR: 1.11, 95% CI: 1.09 to 1.13, P<0.01), 14-day dialysis (OR: 1.18, 95% CI: 1.14 to 1.23, P<0.01), and 14-day mortality (OR: 1.14, 95% CI: 1.12 to 1.17, P<0.01). In Model 1, which adjusted for age, race, sex, medical admission, hospital type, contrast agent use in the prior 72 hours, and Elixhauser score, the AG at the onset of AKI remained significantly associated 14-day AKI progression (OR: 1.09, 95% CI: 1.07 to 1.11, P<0.01), 14-day dialysis (OR: 1.16, 95% CI: 1.12 to 1.21, P<0.01), and 14-day mortality (OR: 1.19, 95% CI: 1.16 to 1.23, P<0.01). In Model 2, which further adjusted for alert status, ICU admission, blood urea nitrogen (BUN) levels, WBC counts, hemoglobin levels, duration of AKI, nephrology consults within 14 days, and SOFA score, the AG at the onset of AKI was still significantly associated with 14-day AKI progression (OR: 1.07, 95% CI: 1.04 to 1.09, P<0.01), 14-day dialysis (OR: 1.09, 95% CI: 1.04 to 1.15, P<0.01), and 14-day mortality (OR: 1.10, 95% CI: 1.07 to 1.14, P<0.01) (Table 3).

SUBGROUP ANALYSIS OF THE RELATIONSHIP BETWEEN WBC COUNTS AND THE AKI PROGRESSION AND THE PROGNOSIS OF PATIENTS WITH AKI:

Age: In patients aged ≤65 years, an increased AG was significantly associated with higher risks of 14-day AKI progression and 14-day mortality, but no significant association was observed with 14-day dialysis. However, among patients aged >65 years, an increase in AG was significantly associated with all 3 outcomes.

Sex: Both male and female patients exhibited a significant association between increased AG and all outcomes, including 14-day AKI progression, 14-day mortality, and 14-day dialysis.

Hospital Type and ICU Admission Status: Among patients admitted to urban teaching hospitals, increased AG at AKI onset was significantly associated with all outcomes except 14-day dialysis.

Similarly, patients not admitted to the ICU demonstrated a significant association between increased AG and all outcomes, except for 14-day dialysis. However, patients admitted to suburban teaching hospitals and those in the ICU showed a significant association between increased AG and all outcomes.

Alert status: Non-alert status patients displayed a consistent significant association between increased AG and all outcomes. However, in alert status patients, significant associations were found with all outcomes except for 14-day dialysis.

Duration of AKI: Patients with AKI duration of less than 2 days and 2–7 days showed a significant association between increased AG at the onset of AKI and 14-day AKI progression, and 14-day mortality. Patients with AKI duration of more than 7 days did not show a significant association between the AG at the onset of AKI and any of the outcomes.

Elixhauser Score: Across the majority of the Elixhauser score subgroups, all showed a significant correlation between increasing AG at the onset of AKI and all outcomes, with the exception of an Elixhauser score <5 and ≥10, which showed no association with 14-day dialysis in Elixhauser score <5 and with 14-day AKI progression in Elixhauser score ≥10 (Table 4).

THE RESULTS OF CURVE FITTING AND THRESHOLD EFFECT ANALYSIS:

Curve fitting analysis revealed that the AG at the onset of AKI was positively correlated with 14-day AKI progression, 14-day dialysis, and 14-day mortality. Further threshold effect analysis found that for patients with an AG at the onset of AKI ≤10 mmol/L, there was no significant association between the AG at the onset of AKI and 14-day AKI progression (OR: 1.02, 95% CI: 0.94 to 1.11, P=0.67), 14-day dialysis (OR: 1.01, 95% CI: 0.79 to 1.29, P=0.93), and 14-day mortality (OR: 1.03, 95% CI: 0.91 to 1.16, P=0.63). For patients with an AG at the onset of AKI >10 mmol/L, the AG at the onset of AKI had been significantly associated with all outcomes, including 14-day AKI progression (OR: 1.13, 95% CI: 1.10 to 1.16, P<0.01), 14-day dialysis (OR: 1.18, 95% CI: 1.13 to 1.23, P<0.01), and 14-day mortality (OR: 1.18, 95% CI: 1.14 to 1.22, P<0.01). However, the P value for trend in 14-day dialysis was not statistically significant (Figure 2, Table 5).

Discussion

STRENGTH OF THE STUDY:

The study utilized a larger sample size and a broader population, beyond just critically ill patients, to enhance the representativeness of the results. The study utilized the AG at the onset of AKI, clearly establishing the detection time point of AG. This addressed the limitations of previous studies that relied on the time of admission or lacked clear measurement, enhancing its clinical practicality. The study incorporated comprehensive endpoint indicators to better reflect the impact of AKI on patients and provide more clinically instructive results. The study discovered a nonlinear correlation between AG at the onset of AKI and 14-day AKI progression, and 14-day mortality. Additionally, it found a linear correlation with 14-day AKI progression, enhancing its usefulness in AKI management.

LIMITATIONS OF THE STUDY:

The study was conducted retrospectively, and future studies are required to confirm the findings. The study focused on inpatients, and it is important to confirm if the findings can be applied to a broader group of individuals, including outpatients. The lack of detailed information on specific AKI etiologies in our primary dataset, which prevented a comprehensive analysis of the relationship between acidosis and different AKI etiologies.

Conclusions

An increase of AG at the onset of AKI was associated with an increased risk of 14-day AKI progression, 14-day dialysis, and 14-day mortality. Especially when the AG exceeded 10 mmol/L, the 14-day AKI progression, and 14-day mortality were significantly increased, the AG showed a linear correlation with 14-day dialysis.