Galectin-3 and Soluble CD146 Identify Cardiorenal Injuries in Severe Burn Patients: A Biomarker-Based Approach

Acute kidney injury (AKI) is a common and serious complication following severe burn injuries [1, 2]. Similarly, myocardial injury (MI) frequently manifests in patients experiencing severe burns [3, 4]. AKI or MI are both associated with poorer outcomes in burn patients [1‒4]. Due to factors such as inflammation, and capillary leak, assessing fluid requirement becomes complex. AKI can manifest up to 7 days after admission and may be associated with hypovolemia, cardiac injuries causing congestion, hemolysis, nephrotoxicity (e.g., hydroxocobolamin…) or inflammatory shock [5‒7]. The pathophysiology behind this phenomenon is driven by intricate interactions among cardiac function, vascular volume regulation, and venous congestion [8, 9]. Galectin-3 (Gal3) is a small protein from the lectin family, identified as a plasma marker for AKI, MI, and cardiorenal disease [10‒15]. Its involvement in these conditions stems from a complex pathophysiology linked to the induction of inflammation and further fibrosis, in both cardiac and renal tissues [16‒21]. Soluble cluster of differentiation 146 (sCD146) is an endothelial junction immunoglobulin recognized for its association with congestion and endothelial activation stemming from cardiac or renal injuries [22‒28]. Interestingly, both Gal3 and sCD146 interact in pathological conditions, promoting inflammation and fibrosis [27, 29‒33]. Their association with cardiorenal injuries suggests their potential role as biomarkers for AKI and/or MI. Within the initial 7 days after admission for a severe burn, understanding the mechanisms driving AKI would significantly enhance therapeutic approaches, potentially involving early fluid restriction, vasopressor therapy [34, 35]. Identifying cardiorenal injury in severe burn patients using a bundle of biomarkers could potentially improve the management strategies for such cases. However, this approach has not been evaluated in severe burn populations. Therefore, this study aims to assess a biomarker-based tool to identify patients at higher risk of cardiorenal injury.

This study is an ancillary prospective observational, proof-of-concept, single-center cohort from the PRONOBURN cohort that includes patients admitted to the burn intensive care unit (BICU) of Saint-Louis Hospital (“Assistance Publique des Hôpitaux de Paris”), with severe burn (a total burn surface area [TBSA] >20%, or a full-thickness burn surface area> 10%). Patients were enrolled between August 2017 and February 2019. This study has been approved by our local Ethic Committee (Comité de protection des personnes IV, Saint-Louis Hospital; Institutional Review Board 00003835, protocol 2013/17NICB). Because the Ethical Committee waived the need for written consent, all patients and/or next of kin will be informed and will orally consent to participate. Exclusion criteria are detailed in online supplementary Data 1 (for all online suppl. material, see https://doi.org/10.1159/000540845).

Data and biological samples were collected within the initial 7 days following admission. The collected data included age, sex, body mass index, TBSA, full-thickness burn surface area, mechanism of injury, and patients’ clinical characteristics. Additionally, assessments encompassed Simplified Acute Physiology Score II (SAPS II) [36], Abbreviated Burn Severity Index (ABSI) [37], Sequential Organ Failure Assessment (SOFA) [38, 39], Unit Burn Standard (UBS) [40], treatments administered during the first 7 days post-admission, need for mechanical ventilation, smoke inhalation, hemodynamic data (extravascular lung water index, cardiac index, cardiac ejection fraction assessed using echocardiography, central venous pressure), daily fluid intake, daily urine output, glomerular filtration rate (calculated using plasma creatinine, urine output, and urine creatinine), need for renal replacement therapy (RRT), need for surgery, occurrence of sepsis during ICU care, daily laboratory values, and in-ICU mortality.

All samples for Gal3 and sCD146 measurements were collected as part of the morning routine biological sampling between 6 and 8:00 a.m. during the initial 7 days following admission. These samples were then stored at −80°C before transfer to the central laboratory for blinded analysis.

For Gal3 measurements, a commercially available chemiluminescence immunoassay using ALINITY I on plasma samples (Abbott Diagnostic, Abbott Park, IL, USA) was conducted blindly and performed in triplicate with the ALINITY I GALECTIN-3 200T (9P6022) kit. The quantification limit ranged between 4 and 114 ng/mL, with the detection limit from 0.9 to 228 ng/mL. Both intra- and inter-assay coefficient of variation, as well as interindividual variability, were reported to be under 3%. The 25th, 50th, 75th, 95th, and 97th percentiles in a healthy population were measured at 12.4, 14.8, 18.5, 25.7, and 27.5 ng/mL, respectively (according to manufacturer datasheet) [41].

For sCD146 measurement, the enzyme immunoassay (ELISA) technique from CY-QUANT ELISA sCD146 (Biocytex, Marseille, France) kit was used blindly and performed in triplicate. The mean sCD146 value in a normal population (61 patients) was reported as 273 +/− 70 ng/mL (as detailed in the manufacturer datasheet). The quantification range in a 1/10 dilution was from 10 to 160 ng/mL (according to manufacturer datasheet).

Patients underwent resuscitation in accordance with the Saint-Louis Hospital BICU resuscitation protocol previously published and detailed in online supplementary Data 2 [8].

CRS was defined as the association of AKI stage 1 or more and MI during the 7 first days 24 h after admission based on high troponin value and variation [42]. The definition of AKI or MI was defined from 24 h to 7 days following admission.

AKI definition was based on Kidney Disease Global Outcome (KDIGO) [43] using serum creatinine (sCr) and urine output, during the first 7 days following admission. Baseline sCr was sCr at admission (sCr_adm) except if estimated glomerular filtration rate (eGFR) at admission was <75 mL/min and minimal sCr level during ICU hospitalization (without RRT or history of chronic kidney disease) was inferior to sCr_adm. In that case, we back calculated baseline sCr from the CKD-EPI equation set to 75 mL/min per 1.73 m² [44].

MI definition was assessed using hypersensitive troponin-T (hsTnT) levels: patients were defined with an MI if hsTnT relative variation between admission value and hsTnT levels in the first 7 days was above 20% or above 40 ng/mL 24 h after admission. This definition was the most applicable in a severe critically ill population as consensual definition is not applicable for patients with severe burn injuries [45, 46].

The primary objective was to assess the occurrence of cardiorenal syndrome (CRS) within the first 7 days. The secondary objective was to identify the association between Gal3 and sCD146 with CRS, AKI, MI, and mortality.

Statistical analysis followed the Strengthening the Reporting of Observational Studies in Epidemiology statement (STROBE) guidelines [47]. Continuous data were presented as mean (standard deviation) for normally distributed variables and compared using a Student’s t test. For non-normally distributed variables, data were expressed as median (first [Q1], third [Q3] quantile) and compared using the Wilcoxon test. Categorical variables were shown as count (percentages) and analyzed using the χ² test or, for multivariate analysis, the Kruskal-Wallis test (adjusted with Dunn’s method). Pairwise comparisons were conducted using Pearson’s χ² test adjusted with the Bonferroni method. For time-dependent measurement comparison, a two-way ANOVA was performed and multiple comparison was corrected with Dunn’s method. The Spearman test was utilized to calculate the correlation ratio R-square, assessing the monotonic relationship between continuous variables displayed on a logarithmic scale.

To explore the association between Gal3 levels and outcomes, receiver operating characteristic curves were employed with admission values for predictive performance. Optimal thresholds were determined based on the best area under the receiver operating characteristic curve, reported with corresponding sensitivity, specificity, and accuracy using Youden index method. Comparison of area under the receiver operating characteristic was conducted using the DeLong test, and confidence intervals were calculated using bootstrap methods. The association between biomarkers and outcomes was evaluated through binomial logistic regression, with estimated values expressed as odds ratios (ORs) (95% CI). All reported probability values are two tailed, and p < 0.05 was considered statistically significant. Analyses were performed using R software (version 4.0; http://www.R-project.org) and sensitivity analysis was performed using PredictABEL package.

A total of 280 patients were admitted to BICU of Saint-Louis Hospital between August 2017 and February 2019. Among these last screened patients, 240 did not meet the admission criteria for this study (they were under 20% of TBSA or were moribund [4 patients]). All characteristics of included patients are detailed in Table 1.

Forty patients met the inclusion criteria and were included in this study. The median age was 52 [33.5, 62.3] years, and 17 (42%) were women. Nine (23%) had chronic hypertension, 1 (3%) had chronic heart failure, 3 (8%) had diabetes, 3 (8%) had pulmonary disease. All patients experienced thermal injuries covering 40 [30, 66]% of the TBSA, with 11 (34%) receiving hydroxocobalamin treatment (an important associated factor with AKI) [48].

Fifteen (38%) patients had CRS, 13 patients (32%) had MI without AKI, 6 patients (15%) experienced AKI without MI during the first 7 days. Among all patients with AKI (AKI without MI or CRS), 4 (10%) were KDIGO stage 1, 7 (18%) were KDIGO stage 2, 10 (25%) were KDIGO stage 3.

Timeline of CRS is reported in Figure 1 indicating that the incidence of CRS was increasing after 7 days following admission, from 7 (17%) at day 1 to 38% at day 7. Correlation of hsTnT and sCr was poor as well as Gal3 and sCD146 with an OR of 0.49, p < 0.001, and 0.26, p < 0.001, respectively (online suppl. Fig. 1).

Twelve patients (30%) died in ICU. Mortality rate was higher in patients with CRS with 9 (23%) patients compared to 2 (5%) patients with AKI alone, 1 (2.5%) patient with MI alone. No patient died with no AKI or no MI. Patients with CRS had a more severe AKI as 8 (53%) had KDIGO, 3 AKI, and most of them were treated with RRT.

The median admission values of Gal3 and sCD146 were 35.7 [29.8, 52.7] ng/mL and 240.5 [152.3, 295.5] ng/mL, respectively. Their kinetic showed that their values were higher at admission and gradually decreased after 7 days following admission (Fig. 2; Table 1).

Gal3 and sCD146 levels were higher in patients exhibiting CRS (Fig. 2). The values are displayed in Table 1. Gal3 value at admission (D0) demonstrated predictive performance for CRS (as no patient was described with CRS at admission), achieving an AUC of 0.78 (95% CI: 0.63–0.93), while sCD146 at admission (D0) showed an AUC of 0.52 (95% CI: 0.32–0.71) (Fig. 3; Table 2). Both biomarkers values during the first 7 days following admission were associated with CRS, presenting an OR of 1.145 (95% CI: 1.081–1.211), p < 0.001, for Gal3 and 1.007 (95% CI: 1.002–1.0143), p < 0.001, for sCD146 (Fig. 4).

The addition of sCD146 at admission (D0) to Gal3 level at admission (D0) improved the prediction of CRS with an AUC of 0.81 (95% CI: 0.66–0.96) (Fig. 3; Table 2). Combining sCD146 with Gal3 pooled values during the first 7 days following admission improved the association with CRS with an OR of 1.147 (95% CI: 1.085, 1.212), p < 0.001 (Fig. 4).

Gal3 pooled values during the 7 days following admission were associated with AKI without MI, displaying an OR of 1.129 [95% CI: 1.065–1.195], p < 0.001, for Gal3 value during the 7 first days following admission (Fig. 4) and a correlation with sCr during the first 7 days following admission (OR = 0.57, p < 0.001) (online suppl. Fig. 2). Correlations with uCr, pUrea, and urine output are displayed in online supplementary Figure 2. sCD146 pooled levels during the first 7 days were not associated with AKI without MI, displaying an OR of 0.994 (95% CI: 0.986–1.003), p = 0.17 (Fig. 4). Gal3 pooled levels during the first 7 days following admission were associated with MI without AKI with an OR = 1.095 (95% CI: 1.037–1.157), p < 0.001 (Fig. 4), and were correlated with hsTnT with an OR of 0.47, p < 0.001 (online suppl. Fig. 3).

sCD146 pooled levels were not associated with MI without AKI, with an OR of 1.002 (95% CI: 0.996–1.007), p = 0.55 (Fig. 3), and were correlated with hsTnT with an OR of 0.13, p < 0.029 (online suppl. Fig. 3). Correlations with plasma BNP, central venous pressure, and extravascular lung water index are displayed in online supplementary Figure 3.

Combining Gal3 to sCD146 pooled value during the 7 first days following admission did not improve association with MI without AKI (Fig. 4). When evaluating specific cardiac failure through visual echocardiographic assessment, Gal3 and sCD146 values were higher within the first 3 days following admission in patients with a low estimated cardiac ejection fraction (online suppl. Fig. 4).

Both biomarkers during the first 7 days following admission slightly exhibited higher levels in non-surviving patients, more specifically for Gal3 (online suppl. Fig. 5; online suppl. Table 1). Gal3 pooled value during the first 7 days following admission demonstrated an association with mortality, displaying an OR of 1.028 (95% CI: 1.012–1.045, p < 0.001), while sCD146 pooled values during the first 7 days were poorly associated with mortality, showing an OR of 1.006 (95% CI: 1.003–1.010, p < 0.001). Combination of pooled biomarkers value during the first 7 days did not improve association with mortality (Fig. 4). Gal3 had a prediction performance for mortality with an AUC of 0.67 (95% CI: 0.47–0.87) and sCD146 an AUC of 0.54 (95% CI: 0.33–0.75) (online suppl. Fig. 5; Table 2).

In this study, CRS is frequent and a severe condition. Gal3, but not sCD146 levels, in severe burn patients, was associated with CRS. Admission values of Gal3 or the combination of Gal3 and sCD146 predicted CRS. Gal3 was also associated with AKI or MI in our cohort and using combination with sCD146 slightly helped to specify patients with CRS.

Identifying patients with CRS poses a challenge as the diagnosis is typically made after clinical damage has occurred [49, 50]. The use of biomarkers for early identification of this syndrome represents a breakthrough in medical management, enabling timely intervention. Within our population of burn patients, assessing cardiorenal injury requires continuous monitoring [51, 52]. Patients with severe burns experience an early hypovolemic phase leading to organ dysfunction due to hypoperfusion, followed by an inflammatory phase resulting in congestion [5]. These phases are associated with kidney or cardiac injury, leading to isolated dysfunction or both cardiorenal dysfunctions [1‒4]. Given the dynamic nature of the patient’s hemodynamic and systemic profiles, we analyzed biomarkers and clinical parameters during the initial 7 days after admission.

Previous studies have evaluated biomarkers for predicting CRS, primarily focusing on specific cardiac or renal biomarkers like KIM-1, CysC, BNP, or troponin [50]. In this study, the choice of Gal3 and sCD146 stemmed from their involvement in the pathophysiology of type 3 CRS [19, 21, 33, 53]. Gal3 is primarily associated with the interplay between the kidney and heart through endothelial and inflammatory mechanisms, leading to cardiac fibrosis and dysfunction in preclinical models [54‒56]. Furthermore, Gal3 is strongly associated with kidney dysfunction and damage [14, 15] and cardiac dysfunction with a low eGFR [18, 57, 58]. This protein is released from kidney tubules and from immune cells that infiltrate kidney and heart after kidney injury [19, 53].

Concerning sCD146, the soluble form of CD146, a cleaved protein of mainly endothelial, was observed to be associated with cardiac congestion and renal disease [24, 33]. Endothelial matters are a major point of organ crosstalk as it linked two organs through systemic mechanisms [59, 60]. Combining Gal3 and sCD146 offers insight into systemic pathophysiology resulting from cardiorenal crosstalk in our study. Moreover, preclinical studies identified specific treatment (as Gal3 inhibitors) to prevent cardiorenal damage that could be applied in our population when biomarkers are increased [19].

Kidney injury was identified using KDIGO criteria, considering both relative and absolute variations in sCr [43]. Baseline sCr was calculated using either back-calculation or sCr at admission, depending on whether the patient had a normal eGFR at admission. This approach minimized misinterpretation regarding AKI for patients with AKI upon admission. Additionally, urine output over 24 h was considered for AKI definition, as it is an important parameter for renal and cardiovascular issues [61], but due to irregular recording of urine collection quantities throughout the days.

Assessment of cardiac injury can rely on echocardiography or could involve plasma biomarker measurements such as BNP and troponin levels [62]. Evaluation of cardiac ejection fraction is not always feasible due to severe skin burns or bandages limiting echogenicity and the increased risk of infectious contamination in patients with compromised skin integrity. This underscores the importance of using biomarkers in this population. However, the increase in these biomarkers could be limited by the severity of the patients’ conditions and the renal function. In critically ill patients, these biomarkers have many limitations due to systemic injuries. The definition of MI remains a challenge in this context. Using the consensual definition (hsTnT >14 ng/mL) [46] would lack specificity in our cohort. Therefore, we used the definition assessed for patients after non-cardiac surgery as they have significant injuries that could reflect the context of severe burns [42, 45]. A specific definition for severe burn patients is needed to fully assess CRS in this population. In a burn population, the massive fluid intake during early management interferes with the concentration of biomarkers as it could be diluted. Biomarkers are often associated with the severity of the patients, reducing their clinical value to predict AKI or cardiac damage as these last appear in severe patients. Interestingly in this study, biomarkers were associated with mortality but not with the severity of the patients defined by the burned surface or the organ dysfunction severity scores.

While utilizing biomarkers to predict cardiorenal damage in the early phase is crucial, their predictive utility beyond ICU discharge would provide more comprehensive insights. However, post-ICU discharge data were unavailable for analysis in our study, and long-term prognosis was not evaluated as organ interaction could be a slow process [63, 64].

Our study demonstrated good predictive performance for AKI using Gal3, and its predictive performance for isolated cardiac damage was also significant. This discrepancy likely emphasizes the importance of AKI prediction using biomarkers in cases of cardiorenal damage, potentially mitigating confusion bias; the study hypothesized that combining biomarker (Gal3 and sCD146) would help to specify CRS. Although this study explored cardiorenal injury, the pathophysiology and epidemiology vary across different types of CRSs [64, 65]. Notably, this study did not differentiate between type 1 CRS and type 3 CRS (type 1 = initial insult is cardiac leading to renal injury, type 3 = initial insult is renal leading to cardiac injury) due to the division of our cohort resulting in small groups, which would compromise validity.

The primary limitations of this study include its retrospective analysis despite prospective data collection, a limited number of included patients due to its proof-of-concept nature, and the need for a point-of-care study to substantiate the relevance of these biomarkers. However, this study is a proof of concept to introduce the importance of identifying patients with cardio-renal injury and more importantly in severely burn patients. To reinforce this result, more large prospective studies are needed, observational and interventional, including biomarker-based protocol assessment to prevent cardiorenal injury.

In severe burn patients, CRS was a frequent and severe condition. Gal3 levels, but not sCD146 levels, were associated with CRS. Admission values of Gal3 alone predicted CRS and enabled better identification of patients with cardiorenal damage. Additionally, Gal3 was also associated with AKI or MI in our cohort, and the combination of Gal3 with sCD146 further improved prediction performance for CRS.

The authors would like to thank the biochemistry department of hôpital Saint for their precious help on this article.

This study has been approved by our local Ethic Committee (Comité de protection des personnes IV, Saint-Louis Hospital; Institutional Review Board 00003835, protocol 2013/17NICB). Because the Ethical Committee waived the need for written consent, all patients will be informed and will orally consent to participate.

L.B., S.S., A.G.L., E.M., B.D., M.B.-C., S.F., C.E.C., and F.A. have no conflict of interest to declare. F.D. received lecture fees from Sedana Medical and bioMerieux and research grant from the French Ministry of Health, European Society of Intensive Care Medicine, and Société Française d’Anesthésie Réanimation. M.L. M.L. reports consulting fees from Novartis, lecture fees from Baxter and Fresenius, and research support from SphingoTec.

This study was supported by a grant from the “Association des Gueules cassées” to Matthieu Legrand and by a grant from “agence nationale de la recherche.”

L.B., S.S., and F.D. designed the study. L.B. analyzed data, made tables and figures, and wrote the manuscript. F.D. supervised the analysis and the writing of the manuscript. L.B., S.S., A.G.L., E.M., B.D., M.L., M.B.-C., S.F., C.E.C., F.A., and F.D. edited and reviewed the manuscript.

The data that support the findings of this study are not publicly available due to their containing information that could compromise the privacy of research participants but are available from the corresponding author [L.B.] upon reasonable request.