Serum sodium trajectory during AKI and mortality risk


 Introduction: Kidneys play a primary role in electrolyte homeostasis. The association between serum sodium level and mortality or the need for kidney replacement therapy during acute kidney injury has not been adequately explored. Methods: In this prospective cohort study, we enrolled patients admitted to the Civil Hospital of Guadalajara from August 2017 to March 2020. We divided patients into five groups based on the serum sodium level trajectories up to ten days following hospitalization, 1) stable normonatremia (serum sodium 135 and 145 mEq/L), 2) fluctuating serum sodium levels (increased/decreased in and out of normonatremia), 3) uncorrected hyponatremia, 4) corrected hyponatremia, and 5) uncorrected hypernatremia. We assessed the association of serum sodium trajectories with mortality and the need for kidney replacement therapy (secondary objective). Results: A total of 288 patients were included. The mean age was 55±18 years, and 175 (60.7%) were male. Acute kidney injury stage 3 was present in 145 (51%). Kidney replacement therapy started in 72 (25%) patients, and 45 (15.6%) died. After adjusting for confounders, 10-day hospital mortality was significantly higher in group 5 (HR, 3.12; 95% CI, 1.05 to 9.24, p = 0.03), and kidney replacement therapy initiation was higher in group 3 (HR, 2.44; 95% CI, 1.04 to 5.70, p = 0.03) compared with group 1. Conclusion: In our prospective cohort, most patients with acute kidney injury had alterations in serum sodium. Uncorrected hypernatremia was associated with death, and uncorrected hyponatremia was correlated with the need for kidney replacement therapy.

ated with substantial morbidity and mortality. Dysnatremias during AKI are common [1]. Thus, it is not surprising that dysnatremias are associated with complications during hospitalization, including the risk of death [2]. Serum sodium (sNa) levels and their alterations during AKI and their relationship with clinical outcomes have been poorly reported. Nevertheless, it is possible that the trajectory of sNa could also be associated with unfavorable clinical outcomes. This study investigated the association between the trajectory of sNa during hospitalization among patients with AKI and their outcomes.

Study design and patient population
In this prospective cohort study, we enrolled patients admitted to the Civil Hospital of Guadalajara Fray Antonio Alcalde, Mexico. We included patients with AKI who received nephrology consultation and were followed for the first ten days after their AKI diagnosis. AKI was diagnosed only by the serum creatinine (sCr) criterion of the KDIGO definition [3]. The exclusion criteria were patients with stage 5 chronic kidney disease (CKD), patients with a hospitalization stay shorter than 48 h, all transplant patients, pregnant patients, and patients with missing data (unable to complete the analysis). The study was approved by the Institutional Review Board and was conducted according to the Declaration of Helsinki. Informed consent was obtained from all the subjects. The protocol followed the STROBE guidelines.

Data collection
The clinical characteristics, demographic information, and laboratory data were collected prospectively. The main predictor was the in-hospital sNa trajectory. For each patient, in-hospital sNa trajectory was assessed based on the sNa values during the hospital stay.

Serum sodium trajectories
The group-based sNa trajectory modeling and the patterns of change for measures across multiple time points were categorized into five main patterns: (1) stable normonatremia, with sNa between 135 and 145 mEq/L; (2) fluctuating sodium, with sNa that increased/decreased in and out of normonatremia; (3) uncorrected hyponatremia, with sNa < 135 mEq/L; (4) corrected hyponatremia, with sNa < 135 mEq/L on hospital admission that increased to normonatremia; and (5) uncorrected hypernatremia, with sNa > 145 mEq/L ( Supplementary Fig. 1). The primary outcome was in-hospital or 10-day follow-up mortality. The secondary outcome was the need for kidney replacement therapy (KRT) initiation.

Statistical analysis
Continuous variables are summarized as the mean ± SD unless otherwise specified. Categorical variables were summarized as numbers with percentages. Survival and KRT initiation were estimated using the Kaplan-Meier plot using the log-rank test. Multivariable hazard model analysis was performed to assess the independent association between in-hospital sNa trajectories and the primary and secondary outcomes. We performed a subgroup analysis using a univariate Cox proportional hazards model to estimate the adjusted hazard ratio (HR) and the associated 95% confidence interval (CI). Patient age, sex, diabetes mellitus, hypertension, admission for surgery, sepsis, CKD, and stage 3 AKI according to the KDIGO criteria were all included in the Cox model.

Subgroup analysis
We performed a subgroup analysis to confirm the increased risk of death in patients with AKI and uncorrected hypernatremia. Patients < 65 years old, males, those with diabetes mellitus, hypertension, admitted by a medical service, sepsis, stage 3 AKI and KRT had worse survival (p for interaction < 0.05 for all).
We then repeated this process to assess the association between uncorrected hyponatremia and the increased requirement for KRT. In patients with uncorrected hyponatremia, the subgroup analysis showed that patients without diabetes mellitus, those admitted to a medical service, and those with stage 3 AKI had a significantly increased risk of KRT initiation during the 10-day follow-up (p value for interaction < 0.05 for all).

Discussion
In this prospective, single-center cohort, we found that > 80% of the AKI patients had fluctuations in sNa. Uncorrected hypernatremia was associated with an increase (> threefold) in the probability of death, even after adjusting for potential confounders. We also found that uncorrected hyponatremia was associated with a higher need for KRT initiation, and these associations were stronger in patients with more comorbidities and a worse stage of AKI.
Dysnatremia and kidney injury may be separate manifestations of a common underlying disease or reflect the severity of a patient's illness and comorbidities [4]. Many pathological conditions may lead to both dysnatremia and AKI, such as changes in the volume status. In our cohort, AKI with uncorrected hypernatremia was associated with a higher incidence of death, which was very similar to the findings of Darmon et al., who showed that hypernatremia was associated with a greater risk of death in AKI patients by more than fourfold [5]. The variability in sNa levels is likely more harmful than the absolute sodium value [6]. The impact of sodium correction speed in hypernatremic patients on clinical outcomes is debated, with scarce evidence [7,8]. Our findings suggest the benefits of earlier hypernatremia correction even in the presence of AKI.
We also found that uncorrected hyponatremia was common among AKI patients who needed KRT. Hyponatremia leads to hypotonicity, which can cause cellular swelling and membrane rupture. Hyponatremia in AKI could represent a specific phenotype seen in sepsis, inflammation, or volume overload. Our demonstrations of the association of AKI requiring KRT in hyponatremic patients are consistent with previous studies [9].
Our subgroup analysis indicated that specific patient populations were more vulnerable to worse outcomes. Patients with more comorbidities had a higher risk of death, which could be due to more limited physiological reserves that allow them to counteract the negative dysnatremia consequences.
Our study has some limitations. The included cohort of patients was relatively small. We were not able to adjust sNa for the serum glucose levels, did not consider the amount of sodium administered in the treatments and lacked urinary output data, or the strategies that could have been used to decrease sNa. Due to the observational nature of the investigation, the causal relationship between dysnatremias and clinical outcomes could not be established. Thus, our investigation serves solely as a hypothesis-generating study. Our article has some strengths. To our knowledge, our study is the first to identify correlations between sNa trajectories in AKI patients who died and the need for KRT initiation.
In conclusion, we found that sNa fluctuations were frequently observed in most patients with AKI during the first ten days of hospitalization. Uncorrected hypernatremia was associated with death and was correlated with the need for KRT initiation. Larger cohort studies and randomized trials are needed to confirm our findings.