Galectin-3 as Prognostic Predictor in Patients with COVID-19 Acute Respiratory Failure

Background. Galectin-3 is β-galactoside-binding lectin with several roles in immune-inammatory response. To date, there is no evidence of Galectin-3 role as a prognostic predictor in COVID-19 disease. The aim of this study is to clarify the prognostic role of Galectin-3 in patients with COVID 19 acute respiratory failure. Methods. We enrolled 156 consecutive patients with COVID-19 disease. Routine laboratory test, arterial blood gas, chest X-ray or Computed Tomography and Galectin-3 dosage were performed. The primary outcome was to assess Galectin-3 predictive power for 30-day mortality. Secondary outcomes were 30-day Intensive Care Unit admission and Acute Respiratory Distress Syndrome stratication according to Galectin-3 dosage. We performed Mann-Whitney U and Kruskal-Wallis tests for continuous variables comparison. Fisher’s exact test or Chi-square test were used for categorical variables analysis. Relationships between Galectin-3, clinical and laboratory data were identied using Spearman analysis. Receiver Operating Characteristic curves estimated Galectin-3 predictive power for the endpoints. With a xed cut-off of 35.3 ng/ml, Kaplan-Meier with Log-Rank test and Cox Regression were performed to assess mortality and Intensive Care Unit admission risk. Results. Galectin-3 correlated with many other prognostic predictors tested in our analysis. Moreover, patients with serum levels of Galectin-3 above 35.3 ng/ml had increased risk for mortality, Intensive Care Unit admission and severe Acute Respiratory Distress Syndrome. Conclusions. Our study demonstrates the role of Galectin-3 as a predictor of mortality, Intensive Care Unit access and ARDS stratication in patients with COVID 19 acute respiratory failure.


Introduction
Since its rst description in 2019, COronaVIrus Disease 19  has shown an intense lung and systemic aggressiveness, albeit a wide spectrum of clinical phenotypes (1). Considering its complexity, numerous factors have been evaluated to identify fragile patients and to stratify their risk of adverse outcomes (2). At this regard, serum in ammatory markers play an important role, as they are a sign of disease worsening and progression (3)(4)(5). Recently, Galectin-3, a β-galactoside-binding lectin, has raised some interest as a potential marker of lung damage and a possible therapeutic target in COVID-19 disease (6,7). Galectin-3 has pleiotropic effects on the immune response: it modulates immune cells lifecycle, angiogenesis and reparative response after lung injury (8, 9). Galectin-3 is highly expressed on broblast, endothelial and epithelial cells (10), as well as on alveolar macrophages as consequence of a lung damage (11,12). During viral infections, Galectin-3 regulates innate immunity (13), modulating cytokines production and release. Furthermore, Galectin-3 can be a binding site for viruses, easing their entry into immune cells (14). In patients with Severe Acute Respiratory Syndrome-COronaVirus 2 (SARS-COV2) infection, a more severe degree of the disease has been proved to be associated with higher blood levels of Galectin-3 and other various cytokines (15)(16)(17). Despite these interesting ndings in scienti c literature, no speci c data are present about the role of Galectin-3 as a prognostic factor in COVID-19 disease. The aim of this study is to clarify the role of Galectin-3 in patients SARS-COV2 infection admitted to our respiratory intensive care unit due to acute respiratory failure.

Methods
In this single-center retrospective observational study, from September 2020 to March 2021 we enrolled 156 consecutive patients admitted to our respiratory intensive care unit of "Policlinico" University Hospital of Bari, Italy, with a diagnosis of COVID-19 disease and acute respiratory failure. At the emergency department, a nose-throat swab with Real Time-Polymerase Chain Reaction has been performed to con rm SARS-COV2 infection. In our ward, blood samples, arterial blood gas analysis and thoracic imaging (chest X-ray or Computed Tomography-scan) were collected within 48 hours from admission.
Similarly, demographic, anamnestic and clinical data were recorded and reported in a database along with serum dosages of Galectin-3 and other in ammation markers. Plasma samples were collected and stored at -80°C before the analysis. Then, plasma levels of Galectin-3 were measured with chemiluminescence immunoassay kits. Exclusion criteria in our study were the follows: age < 18 years, no blood samples collection within 48 hours, no thoracic imaging performed at the admission. Finally, 140 patients met all the inclusion criteria and were considered for statistical analysis. All these patients were also strati ed for Acute Respiratory Distress Syndrome (ARDS) severity according to the Berlin de nition (18). The primary outcome of this study was to assess 30-day mortality according to Galectin-3 serum levels. Secondary outcomes were the assessment of 30-day Intensive Care Unit (ICU) admission and ARDS strati cation. The study was approved by the Institutional Review Board of our hospital (Ethical Committee number: 6717). The present study was conducted in accordance with the Helsinki Declaration of 1975 and following the standards of Good Clinical Practice.

Statistical analysis
We veri ed the non-normal distribution of data with the Shapiro-Wilk test, considering medians and interquartile ranges for statistical purposes. Consequently, Mann-Whitney U test was used to compare continuous variables, whereas Kruskal-Wallis test was performed in our ARDS severity strati cation.
Categorical variables were compared using Fisher's exact test or Chi-square test. Spearman correlation analysis was used to identify relationships between Galectin-3 and other clinical and laboratory data. To estimate the predictive power of Galectin-3 for the outcomes, we carried out Receiver Operating Characteristic (ROC) curves, estimating the area under the curve (AUC) of our predictors. Then, Kaplan-Meier analysis with log-rank test was performed using Galectin-3 to stratify our patients according to different outcomes. Moreover, risk factors for 30-day mortality were assessed using a univariate Cox proportional hazard regression model. Finally, statistically signi cant predictors were used to generate a multivariate model of Cox regression analysis, whose accuracy was tested using a ROC curve. All statistical analysis were realized using SPSS 26.0 (SPSS Inc, Chicago, Ill) and Prism 8.0.1 (Graphpad Software, La Jolla, Calif). A p-value level < 0.05 was considered to be statistically signi cant.

Population analysis
Anamnestic, clinical and laboratory characteristics of our population are described in Additional File 1. In our cohort, 95% of patients had ARDS according to Berlin de nition. During the hospitalization, 12 patients underwent a worsening of their respiratory condition and were transferred in ICU.

Statistical analysis for ICU admission
Concerning ICU admission, we considered only 108 patients, as for the rest of them there were contraindications for endotracheal intubation and invasive ventilation in ICU. Among these patients, 12 were transferred in ICU due to worsening of gas exchange. To predict ICU admission using serum levels of Galectin-3, a ROC curve was performed, obtaining an AUC of 0.7 (p = 0.02). With the same cut-off of 35.3 ng/ml xed for mortality analysis, we found a statistically signi cant result concerning risk of ICU admission after Log-Rank test (Fig. 4, χ²=6.5, p = 0.01). Subsequently, a univariate Cox regression was performed to evaluate hazard ratios for ICU access (Table 4) Galectin-3 (HR = 1.037, p < 0.0001) were associated with an increased risk for this outcome. However, after adjusting for confounding factors, none of these parameters remained statistically signi cant (Table 5). Paramethers HR# IC 95%** P value ‡ ‡ CPK, Calcium-dependent Protein Kinase, § NT-pro-BNP, N-Terminal pro-Brain-type Natriuretic Peptide; ll SOFA, Sequential Organ Failure Assessment.  In this case, using the xed cut-off of 35.3 ng/ml, we found a sensitivity of 70.6% and a speci city of 78% for the outcome. Interestingly, similar results were found in Kaplan-Meier analysis (Fig. 5, χ²=20.51, p < 0.0001).Whereas no differences were found between mild and moderate ARDS (see Additional File 9 in the Online Data Supplement, χ²=0.19, p = 0.7), severe ARDS was signi cantly strati ed according to our xed cut-off (Fig. 6, χ²=19.8, p < 0.0001).

Discussion
This is the rst study in scienti c literature assessing the prognostic role of Galectin-3 in acute respiratory failure secondary to COVID-19 disease. Patients with higher serum levels of Galectin-3 tend to develop a more severe degree of ARDS with a worse prognosis. It is well known that SARS-COV2 infection can lead to the so called "cytokine storm" in some susceptible patients (19). For instance, our non-survivors group shows increased blood levels of various in ammation markers, which are frequently associated with negative outcomes in COVID-19 disease (2,20). Nevertheless, only IL-6, CRP and Galectin-3 remain statistically signi cant in our multivariate regression model. This nding is not surprising for IL-6 and CRP, which were previously reported as important prognostic markers in COVID-19 disease (21)(22)(23). On the contrary, this is the rst study addressing this role for Galectin-3. Furthermore, among the explored parameters, Galectin-3 shows the best AUC curve in ROC analysis, suggesting a better predictive power for mortality outcome. As stated before, hyperin ammation can trigger Galectin-3 release from a wide range of host cells (24). Furthermore, increased blood concentrations of Galectin-3 have also been described in heart failure and chronic kidney disease (25,26). Although having higher serum levels in our non-survivor group, both NT-pro-BNP and creatinine did not predict a higher death risk in our multivariate models. For this reason, we speculate that during SARS-COV2 infection, Galectin-3 release can be speci cally associated with lung damage, earlier predicting the clinical worsening of this disease. Indeed, our analysis on ARDS strati cation seems to con rm this hypothesis, as Galectin-3 can properly identify severe ARDS secondary to COVID-19 disease with good sensitivity and speci city. Similarly, Xu el al have already explained the role of Galectin-3 as a prognostic factor in ARDS (27). However, this study was not performed during COVID-19 pandemic, taking into consideration patients suffering from severe pneumonia, burns, aspiration or gastrointestinal lesions. Moreover, unlike our study, the ROC analysis was performed considering a combined model with Galectin-3, Acute Physiology And Chronic Health Evaluation II (APACHE II) score and PaO2/FiO2, in order to gain a su cient predictive power for the outcome.
Concerning the ICU admission, our multivariate analysis did not nd any signi cant predictor for this outcome. It has to be said that our cohort of patients undergoing ICU transfer was very limited. Since our ward was deputed to manage patients requiring non-invasive ventilation, ICU admission was allowed only for patients requiring endotracheal intubation or extracorporeal membrane oxygenation (ECMO). For this reason, 32 patients were excluded from this analysis, as they were considered neither eligible for these treatments nor for ICU admission. By contrast, Kaplan-Meier and ROC analysis did result in a statistically signi cant assessment of risk for ICU admission using our Galectin-3 cut-off. For this reason, we believe that a larger cohort of patients should elucidates this point. For mortality, ICU admission and ARDS severity assessments, we decided to use the same cut-of value of 35.3 ng/ml, as it guarantees us the best compromise in terms of sensitivity and speci city. Furthermore, this cut-off only discriminates severe ARDS from mild-moderate ones and can be really considered in prognostic terms the "edge of the cliff".
Patients with Galectin-3 serum levels above 35.3 ng/ml, in fact, were not only more prone to develop severe ARDS, but also markedly at higher risk of ICU admission or death. Our study has several limitations. Firstly, as previously mentioned, the sample size is limited for a complete statistical analysis. Secondly, we performed a single-center retrospective study, while a randomized multicenter prospective design would be advisable to obtain further prognostic information. Thirdly, we only assessed serum levels of Galectin-3 at the admission, without monitoring its plasma concentrations during the hospitalization. Another important limitation is the lack of a non-COVID ARDS control group. Although no important differences seem to characterize this two types of ARDS (28,29), many pathophysiological aspects have to be clari ed and a direct confront of Galectin-3 serum levels in these two groups could be interesting to better understand this point. Finally, since our study had only prognostic purposes, we did not collect any brochoalveolar lavage or pulmonary tissue sample for Galectin-3 detection. Future studies should verify how Galectin-3 concentrations in these samples could affect COVID-19 ARDS development and prognosis.

Conclusions
Our study demonstrates the importance of Galectin-3 as a prognostic factor in COVID-19 disease.
Galectin-3 can predict mortality and ARDS severity of patients with ARF secondary to COVID-19 disease.
Moreover, higher levels of this marker seem to be correlated with an increased risk for ICU admission, although further studies are needed to clarify this issue.

Declarations
Availability of data and materials The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.