Predictive Value of Serum Soluble ST2 in Adult Patients Undergoing Cardiac Surgery for Acute Kidney Injury

Acute kidney injury (AKI) accompanying cardiac surgery is considered as a well-recognized and serious complication, with a reported up to 42% incidence rate [1, 2]. Cardiac surgery-associated AKI (CSA-AKI) increases the length of intensive care unit (ICU) and hospital stay and total treatment costs [3, 4]. Moreover, AKI patients have a higher risk of chronic kidney disease (CKD) progression, which may ultimately require dialysis or kidney transplantation [5, 6]. AKI is associated with increased long-term mortality after cardiac surgery [7]. Therefore, early identification of AKI may allow for timely intervention to alleviate progression and potentially improve outcomes.

Although demographic and clinical data can alert clinicians to patients who are prone to AKI, the additional use of specific biomarkers might further enhance our ability to identify these patients [8, 9]. It is well recognized that AKI is associated with injury pathways such as inflammation and mechanical factors [10]. In this context, serum soluble ST2 (sST2), as a myocardial protein induced by biomechanical stress and involved in various inflammatory pathways of kidney injury, might become a potential biomarker for identifying CSA-AKI [11, 12].

Serum sST2, an interleukin-1 (IL-1) receptor family member, binds to IL-33 as a decoy receptor and neutralizes the beneficial effects of IL-33 [13, 14]. Recently, the relevant prognostic value of sST2 in various cardiovascular events has been recognized, including heart failure (HF), myocardial infarction, acute aortic dissection, and atrial fibrillation [15‒18]. In addition, sST2 may have an adverse effect on kidney function. Studies have found a significant association between higher levels of sST2 and the progression of CKD [19, 20]. Among patients with myocardial infarction, sST2 can serve as an early predictive indicator of AKI [21, 22]. Moreover, preoperative sST2 levels were higher in postoperative AKI patients undergoing coronary artery bypass graft surgery (CABG) and significantly associated with AKI severity [23]. However, it is not yet clear whether serum sST2 levels at ICU admission have an effect on the risk of AKI after cardiac surgery. The current study seeks to investigate the relationship between serum sST2 at ICU admission and postoperative AKI development in adults undergoing cardiac surgery.

This was a prospective, observational study of patients admitted to the ICU of Guangdong Provincial People’s Hospital. We consecutively included all adult patients (≥18 years old) who underwent cardiac surgery with cardiopulmonary bypass between September 2022 and March 2023. The exclusion criteria consisted of end-stage kidney disease, preexisting dialysis before admission, prior renal transplant, nephrectomy, pregnancy status, lack of admission data, or refusal of consent. Patients’ clinical data were collected from medical records.

From the electronic medical record system, the baseline characteristics of enrolled patients were collected. All variables were as follows: demographic features including age, sex, and weight; preexisting clinical conditions including hypertension, diabetes, coronary heart disease, percutaneous coronary intervention, heart failure, atrial fibrillation, cerebrovascular diseases, hyperlipidemia, previous cardiac surgery, and liver disease; preoperative use of contrast agents; left ventricular end-diastolic diameter and left ventricular ejection fraction; preoperative laboratory tests including hemoglobin, hematocrit, albumin, D-dimer, baseline serum creatinine (sCr), and baseline estimated glomerular filtration rate (eGFR) calculated by the CKD-Epidemiology Collaboration (CKD-EPI) equation [24]; New York Heart Association (NYHA) classification; American Society of Anesthesiologists (ASA) classification; and operative information including type of surgery, emergent surgery, cardiopulmonary bypass time, the volume of blood loss, and intraoperative transfusion, such as red blood cells, platelets, and fresh frozen plasma. Serum sST2 was calculated immediately at ICU admission after cardiac surgery. The Acute Physiology and Chronic Health Evaluation (APACHE) score was assessed after admission to the ICU. The development of AKI was the primary outcome variable. Secondary outcomes included utilization of renal replacement therapy (RRT) or extracorporeal membrane oxygenation (ECMO), mechanical ventilation duration, length of ICU stay, ICU mortality, length of hospital stay, and total hospitalization expenses.

Blood was clotted at room temperature after collecting in serum tubes. Samples were centrifuged at 3,000 rpm for 15 min, and then a pipette was used to separate the serum portion. Serum sST2 was quantified using a microfluidic immunofluorescence assay (Stimulation Expressed Gene 2 Test Cartridge; Mircopoint, China), with an upper limit of detection of 200 ng/mL. The intra-assay and interassay coefficients of variation of sST2 were both ≤15%. The people in charge of measuring sST2 were unaware of patients’ clinical characteristics.

The primary outcome of this study was postoperative AKI within 7 days after admission to the ICU. AKI was scored with the Kidney Disease: Improving Global Outcomes (KDIGO) classification [25], where stage 1 was defined as a 26.5 μmol/L increase or a 1.5–1.9 times the baseline level increase in postoperative sCr, stage 2 as postoperative sCr being a 2.0–2.9 times the baseline level increase, and stage 3 as postoperative sCr a 3 times the baseline level increase or a 353.6 μmol/L increase or the initiation of dialysis. Stage 2 and stage 3 were considered severe AKI. The most recent sCr before ICU admission was determined as baseline sCr. Urine volume was not considered in this study.

Patients were classified into two groups based on whether AKI occurs after cardiac surgery, and the characteristics and clinical outcomes of AKI and non-AKI groups were analyzed. Categorical variables were reported as numbers and percentages. Continuous variables, all non-normally distributed in this study, were presented as the median and interquartile range. For continuous variables, group comparison was using the Mann-Whitney U test or Kruskal-Wallis test, while the χ² test was for categorical variables. To explore the relationship between the sST2 levels and outcomes, we divided the cohort into tertiles based on serum sST2 measurements. The association between sST2 levels and the occurrence of AKI was accessed by multivariable logistic regressions. Baseline characteristics in the univariate analyses with statistical significance (p < 0.1) were entered into the forward-stepwise logistic regression to filter out the independent predictors of AKI.

A basic model was established by predictors except for sST2. To evaluate whether sST2 can improve the accuracy of predicting AKI, C-index, net reclassification improvement (NRI), and integrated discrimination improvement (IDI) were calculated after incorporating sST2 into the basic model [26]. The De Long test was aimed at comparing the C-index of the basic model and the basic model plus sST2 [27]. The NRI indicated how much improvement the new model had made in correctly predicting AKI probability after adding new variables, and the IDI reflected the overall improvement in AKI prediction probability of new model. A p value <0.05 was considered statistically significant.

We enrolled 500 patients aged from 22 to 79 years whose characteristics and clinical outcomes were summarized in Table 1. In total, 207 cases of AKI occurred after cardiac surgery, with a 41% incidence rate. Patients developing AKI were older and more male than non-AKI patients. Hypertension, diabetes, and emergent surgery were more prevalent in the AKI group. Compared with non-AKI patients, the baseline eGFR tended to be lower in AKI patients. Moreover, AKI patients experienced greater volume of blood loss and more intraoperative transfusions of red blood cells. The APACHE scores and sST2 levels tended to be higher in the AKI group. In addition, the AKI group was more likely to develop adverse outcomes, including a higher utilization rate of RRT and ECMO, longer mechanical ventilation time, longer length of ICU and hospital stay, higher ICU mortality, and higher total hospital costs.

According to the serum sST2 measured at ICU admission, patients were stratified into tertiles. Patient characteristics according to ST2 tertiles are summarized in online supplementary Table S1 (for all online suppl. material, see https://doi.org/10.1159/000540529). Figure 1 shows that as the sST2 tertiles progressively increased, the incidence of AKI progressively increased. The incidence of AKI in the lowest tertile of sST2 is 20%, while that in the middle tertile is 34%, and that in the highest tertile is 70%.

Multivariable logistic regression was used to investigate the associations between serum sST2 measured at ICU admission and AKI (Table 2). There is a significant association between elevated serum sST2 levels and increased odds of AKI (adjusted OR 1.008, 95% CI: 1.003–1.013, p = 0.003). When serum sST2 measurements were stratified into tertiles, the middle tertile (adjusted OR 1.969, 95% CI: 1.125–3.447, p = 0.018) and the highest tertile (adjusted OR 4.267, 95% CI: 2.361–7.711, p < 0.001) groups had adjusted ORs for AKI that were gradually larger as gradually higher sST2 levels, using the lowest tertile as the reference group.

Among secondary outcomes, patients with the highest tertile of sST2 showed a trend toward worsening versus patients with lower tertiles (Table 3). In the highest sST2 tertile, the incidence of severe AKI was higher (p < 0.001). In addition, patients with the highest sST2 tertiles were more likely to have higher utilization of RRT (p < 0.001) and ECMO (p = 0.001), higher ICU mortality rates (p < 0.001), and higher total hospital costs (p < 0.001) compared to those with lower tertiles. Moreover, mechanical ventilation duration and length of ICU and hospital stay were longer in the highest tertile of sST2 (p < 0.001).

Table 4 shows the independent predictors of AKI selected through logistic regression analysis. All 26 candidate variables listed in the Table 2 footnotes were entered into the logistic regression, and only the 8 best predictors were selected: sex, preexisting diabetes, baseline eGFR, emergent surgery, the volume of intraoperative blood loss, intraoperative RBC transfusion, APACHE scores, and sST2 levels at ICU admission.

Then, we aimed to explore the predictive value of adding sST2 into a basic model constructed by selected predictors other than sST2. Three ROC curves of sST2, basic model, and basic model plus sST2 were displayed in Figure 2. The cutoff point of sST2 that best predicted the development of AKI was 49.71 ng/mL. As shown in Table 5, adding sST2 to the basic model could improve the prediction for AKI slightly from C-index 0.822 to 0.840, but significantly (p= 0.031). Furthermore, adding sST2 into the basic model could significantly improve the reclassification for AKI compared to the basic model alone (p < 0.001). Similarly, the IDI showed an improvement in AKI prediction when adding sST2 (p < 0.001).

When excluding patients undergoing emergent surgery, the associations between higher sST2 and AKI remained significant. The results were concordant with the main analysis (online suppl. Table S2).

In this study, the association between serum sST2 levels at ICU admission and postoperative AKI in adults undergoing cardiac surgery was investigated. Patients with higher sST2 levels had a higher incidence of AKI. Meanwhile, as progressively higher tertiles of serum sST2 levels increased, the ORs for AKI progressively increased. Even after adjusting for confounding factors, the relationship was still maintained. Furthermore, sST2 can enhance the prediction of AKI compared to the basic model. Therefore, sST2 may be a useful biomarker in the early identification of AKI in clinical practice.

The application of serum biomarkers is an additional and promising method to identify AKI early [10, 28]. sST2 is not significantly influenced by age or body mass index, which gives sST2 a practical advantage for predicting AKI [29]. As a novel cardiac biomarker, sST2 had higher specificity for acute HF than B-type natriuretic peptide [30]. Interestingly, natriuretic peptide is considered a predictor of AKI after cardiac surgery [31, 32]. These studies prompt us to speculate that sST2 might enhance the early prediction of AKI after cardiac surgery.

The possibility of serum sST2 affecting renal function has been previously investigated in CKD. An elevated concentration of serum sST2 has been found in CKD patients [33]. In addition, among patients with CKD, elevated levels of sST2 were significantly associated with CKD progression and adverse clinical outcomes [19]. The role of sST2 in AKI has been explored in few studies. Rises of the sST2 level have prognostic power in predicting AKI after myocardial infarction [21]. In another study, preoperative sST2 was elevated in postoperative AKI patients undergoing CABG [23]. However, these studies did not explore certain known confounding factors that affect the development of AKI, such as baseline eGFR, intraoperative blood loss, and transfusion. Therefore, in the multivariate analysis of our study, we confirmed a significant correlation between AKI and multiple perioperative clinical factors, such as sex, preexisting diabetes, baseline eGFR, emergent surgery, volume of blood loss, intraoperative RBC transfusion, and APACHE scores. In previous studies, these risk factors have been reported [34‒39]. Overall, sST2 levels at ICU admission were significantly associated with the development of postoperative AKI even after adjusting for confounding variables. The C-index of sST2 at ICU admission was 0.757 for postoperative AKI, which was potentially helpful. Although the addition of sST2 improved the prediction for AKI slightly beyond the basic model, the NRI was 0.390 (0.217, 0.562), indicating that the risk classification was improved by 39%. Therefore, the data presented in this study not only support sST2 as an independent predictor of AKI but also show that it assists in improving the early prediction of AKI. Further potential treatments or actions could be implemented in cases where elevated ST2 levels indicate a high risk of developing AKI. These interventions aim to prevent AKI and reduce the likelihood of future kidney failure.

The pathophysiology underlying the association of sST2 with CSA-AKI has yet to be fully clarified. As a decoy receptor, sST2 can neutralize the effect of IL-33, which has been found to be the ligand of sST2 [40]. IL-33 exerts cardioprotection by reducing cardiomyocyte apoptosis, fibrosis, and hypertrophy, while sST2 partially inhibits these effects [41‒44]. In addition, research has indicated that IL-33 treatment transforms the inflammatory milieu after kidney injury into a type 2 response, which can ameliorate the course of progressive, proteinuria CKD induced by adriamycin [45]. IL-33 treatment can lead to more efficient bacterial clearance and alleviate systemic proinflammatory responses, while individuals who have not recovered from sepsis have significantly more sST2 than those who have recovered [46]. Moreover, in mice subjected to renal ischemia-reperfusion injury, IL-33 significantly alleviates renal injury and reduces mortality [47]. According to other reports, IL-33 exacerbates cisplatin-induced and ovalbumin-induced AKI [48, 49]. Considering the role of sST2 and IL-33 in the heart, kidney, and inflammatory response, the action of sST2 is likely to be determined by the microenvironment, specific cells, and various AKI models. Therefore, due to limited evidence on the association between sST2 and kidney injury, further investigations are needed to illuminate the underlying mechanisms involved.

Our study also has limitations. First, it was conducted at a single institution, and relatively few patients were included. Second, the association between sST2 and the outcome of mortality was not explored because there were only 16 deaths in this study. Third, we only defined AKI on the basis of changes in serum creatinine, as urine volume cannot be strictly and accurately recorded in clinical practice, as well as the potential alterations of medical therapy on urine output. This may lead to neglecting a portion of AKI patients, as determined by urine output. In addition, despite several predictors of AKI have been assessed in the multivariable analysis, information on hemodynamic status and relevant biomarkers such as natriuretic peptides and troponin was lacking, which could influence the conclusions. Finally, we only examined serum sST2 at the point of ICU admission and did not dynamically observe its changes. Thus, future studies should be conducted to continue validating the relationship between sST2 and AKI.

In summary, patients developing AKI after cardiac surgery appear to have elevated serum sST2 levels at ICU admission compared to those without AKI. The results presented in this study underpin that higher sST2 levels are independently associated with the development of AKI.

The authors would like to thank all the doctors, nurses, technicians, and patients participating in this study for their dedication.

Following the Declaration of Helsinki, this study was approved by the Ethics Committee of the Guangdong Provincial People’s Hospital (Registration No.: KY2020-103-01). All patients signed written informed consent.

The authors have no conflicts of interest to declare.

This work was supported by the National Natural Science Foundation of China (8217080738 to Chunbo Chen).

Zeling Chen and Chunbo Chen contributed to the study design and concept. Chunbo Chen and Yiyu Deng helped conduct the study and took responsibility for the integrity and accuracy of the data. Jiaxin Li and Xicheng Liu performed statistical analysis. Zeling Chen, Jiaxin Li, and Xicheng Liu drafted the manuscript. Xiaolong Liu, Junjiang Zhu, and Xuanhe Tang helped collect data and review the analysis of the data.

Zeling Chen, Jiaxin Li and Xicheng Liu contributed equally to this work.

The data that support the findings of this study are not publicly available for protecting privacy of research patients but are available from the corresponding author [C.B. C.] upon reasonable request.