Galectins can serve as biomarkers in COVID-19: A comprehensive systematic review and meta-analysis

Background Galectins are an eleven-member class of lectins in humans that function as immune response mediators and aberrancies in their expression are commonly associated with immunological diseases. Several studies have focused on galectins as they may represent an important biomarker and a therapeutic target in the fight against COVID-19. This systematic review and meta-analysis examined the usefulness of clinical assessment of circulating galectin levels in patients with COVID-19. Methods International databases including PubMed, Scopus, Web of Science, and Embase were systematically used as data sources for our analyses. The random-effect model was implemented to calculate the standardized mean difference (SMD) and a 95% confidence interval (CI). Results A total of 18 studies, comprising 2,765 individuals, were identified and used in our analyses. We found that Gal-3 is the most widely investigated galectin in COVID-19. Three studies reported significantly higher Gal-1 levels in COVID-19 patients. Meta-analysis revealed that patients with COVID-19 had statistically higher levels of Gal-3 compared with healthy controls (SMD 0.53, 95% CI 0.10 to 0.96, P=0.02). However, there was no significant difference between severe and non-severe cases (SMD 0.45, 95% CI -0.17 to 1.07, P=0.15). While one study supports lower levels of Gal-8 in COVID-19, Gal-9 was measured to be higher in patients and more severe cases. Conclusion Our study supports Gal-3 as a valuable non-invasive biomarker for the diagnosis and/or prognosis of COVID-19. Moreover, based on the evidence provided here, more studies are needed to confirm a similar diagnostic and prognostic role for Gal-1, -8, and -9.


Introduction
In December 2019, the coronavirus (CoV) disease of 2019  first appeared in Wuhan, Hubei Province (1). There are major diagnostic and assessment challenges associated with the COVID-19 pandemic brought on by severe acute respiratory syndrome (SARS) CoV-2 (SARS-CoV2) as the pathogen responsible for the disease. These include figuring out the importance of signs and symptoms in predicting potential infections, determining whether current biochemical and laboratory tests can detect infections and identify patients needing immediate medical attention, and seeing if new diagnostic tests can enable precise rapid and point-of-care testing (2).
In this regard, biomarkers play a crucial role in the clinical decision since they can anticipate the COVID-19 severity and track its prognosis (3). For this reason, in the last few years, several biomarkers have been investigated and identified. For instance, studies have demonstrated that in patients with a higher severity level of COVID-19 pneumonia, increased serum levels of C-reactive protein, interleukin-10, and interleukin-6/Interferon-g ratio were frequently found (4)(5)(6). In addition, others reported that blood levels of D-dimer, lactate dehydrogenase, and white blood cell count played a role in predicting adverse outcomes in the severe COVID-19 (7).
Galectins are a group of lectins with a phylogenetically conserved structure that share approximately 130 amino acids that make up the consensus amino acid sequence, and the carbohydrate recognition domain is in charge of binding bgalactoside (8,9). Galectin family members have been identified in circulating cells, main and secondary lymphoid tissues, and other organs (10). To date, fifteen mammalian galectins have been discovered and only eleven are found in humans (11). Among these, galectin (Gal)-1, -3, -8, and -9 have been well-described as key immune response mediators. While Gal-1 generally exhibits suppressive activities on the immune system, Gal-3 and -9 have been shown to suppress or amplify inflammatory responses (12). Likewise, Gal-8 has also been suggested to have disparate roles in immunity, functioning as an enhancer of the expression and secretion of many pro-inflammatory cytokines or as a cytokinebinding partner to sequester and hinder cytokine activity (12). For this reason, galectins have been linked to a variety of inflammatorydriven disorders (e.g., cancer, cardiovascular, and metabolic diseases) (11)(12)(13)(14)(15)(16). Moreover, since these molecules are also involved in pro-fibrotic processes (13), their role is particularly relevant because they are involved in the pathogenesis of several disorders affecting several organs including the heart and lungs (16)(17)(18).
Based on this evidence, recent studies suggest the clinical utility of evaluating Gal-1, -3, -8, and -9 circulating levels as novel potential diagnostic and prognostic biomarkers of COVID-19 either alone or in conjunction with other known biomarkers (19)(20)(21)(22). Hence, here we conducted a systematic review and a metaanalysis to combine the findings from previous studies assessing Gal-1, -3, -8, and -9 as putative, non-invasive biomarkers of COVID-19 severity and COVID-19-related complications (e.g., cardiovascular). The conclusions obtained by this study can pave the way for future research on these biomarkers and further enlighten their role in clinical settings.

Inclusion and exclusion criteria
Inclusion criteria were clinical studies that (1): compared galectin levels between COVID-19 patients and healthy controls (2), compared galectin levels between different severities of COVID-19 (3), compared galectin levels between patients with comorbidity and without comorbidity related to COVID-19, or (4) used galectin levels to discriminate COVID-19 patients from healthy controls or differentiate different severities of COVID-19. We excluded animal studies, case reports, reviews, and conference abstracts. Finally, non-English abstracts were also excluded.

Screening, data extraction, and quality assessment
Required data were extracted by two reviewers (AHB and SYA) in a pre-designed sheet by the third reviewer (AKh). We extracted these for each study (1): study characteristics including publication year, locations(s), and first author's name (2), population characteristics including total population number, mean age, and female percentage, and (3) main findings.
The Newcastle-Ottawa Scale (NOS) for non-randomized studies was used to assess the risk of bias in included studies (23). NOS evaluates studies according to pre-specified items including selection, comparability, and exposure. Studies with a total NOS ≥ 8 are considered high quality.

Statistical analysis and data synthesis
Data synthesis was performed from the mean and standard deviations (SDs) of galectin levels in COVID-19 cases or different severities of the disease. As there were sufficient studies for Gal-3, we conducted meta-analyses for comparison of this biomarker between COVID-19 cases vs controls, and severe vs. non-severe forms of the disease. Random-effect meta-analysis was performed to calculate the pooled effect size using standardized mean difference (SMD) and 95% confidence interval (CI). To find the source of heterogeneity, meta-regression was conducted with variables including sample size, mean age, and female percentage. The associated bubble plots were also shown. Moreover, publication bias was evaluated using the visual symmetry of funnel plots and statistical tests of Begg's and Egger's (24,25). Finally, to find the possible outliers of analyses, the Galbraith plots were designed, and sensitivity analysis was performed with the removal of each study in analyses.

Gal-1 levels comparison between COVID-19 patients and controls
The association between Gal-1 levels and COVID-19 was evaluated by three studies (20,22,35). In a case-control study by De Biasi et al. (20), Gal-1 levels were compared between 39 COVID-19 patients and 25 age-and sex-matched healthy controls. They found significantly higher levels of Gal-1 in COVID-19 patients compared to healthy controls. In line with this study, Kazancioglu et al. (22) found higher serum levels of Gal-1 in COVID-19 patients compared to healthy controls (P < 0.001).

Gal-1 levels in different severities of COVID-19
The study by Kazancioglu et al. (22) found no significant difference between severe (n = 29) and non-severe (n = 55) COVID-19 patients (median = 9.86 vs. 6.35 ng/mL). In contrast, Markovic et al. (35) found higher levels in stage III of COVID-19 compared to stages I and II (P = 0.001). Detection of stage III COVID-19 was possible with a cut-off of 9359 pg/mL (specificity: 25.9% and sensitivity: 72.2%). In addition, stage II was also associated with higher levels of Gal-1 compared to stage I (P < 0.05) and detection of stage II COVID-19 was possible with a cut-off of 3959 pg/mL (specificity: 23.4% and sensitivity: 89.5%).
Meta-regression was performed to find the source of heterogeneity for this analysis, and it was shown that mean age had a significant association with the result observed (P = 0.041). In addition, mean age contributed to the 44.05% of the heterogeneity observed. Sample size and female percentage did not have any association with the Gal-3 levels in COVID-19 cases ( Table 2). The bubble plots for meta-regression based on sample size, mean age, and female percentage are shown in Supplementary Figures 2-4.
The funnel plot for the assessment of publication bias is shown in Supplementary Figure 5, which showed apparent asymmetry and introduced the possibility of publication bias. However, Begg's and Egger's statistical tests did not show any significant publication bias (P = 0.367 and P = 0.186, respectively).

Meta-analysis of Gal-3 levels in Severe vs. non-severe COVID-19
Five studies reported Gal-3 blood concentrations in severe COVID-19 cases and compared it with non-severe patients. Meta-analysis for these levels showed a tendency toward higher levels of Gal-3 in severe instances of COVID-19; however, this was not significant (SMD 0.45, 95% CI -0.17 to 1.07, P = 0.15, Figure 4). High heterogeneity was observed for this analysis as well (I 2 = 91.63%). The Galbraith plot did not reveal any outlier study for this analysis (Supplementary Figure 6). However, in sensitivity analysis, removing Karsli et al. made the overall result significant (33).
In performing meta-regression for this meta-analysis, the sample size was found to have a direct association with the  Forest plot for the meta-analysis of Gal-3 levels in COVID-19 patients vs. healthy controls.
pooled effect size (P = 0.041). In addition, sample size and female percentage accounted for 48.90% and 3.53% of the observed heterogeneity, respectively ( Table 2, bubble plots in Supplementary Figures 7-9). Publication bias investigation based on the funnel plot inspection did not show any sign of bias (Supplementary Figure 10). In line, Begg's and Egger's statistical tests were not significant for publication bias (P = 0.227 and P = 0.178, respectively).

Gal-3 levels' association with COVID-19 complications
Intensive care unit (ICU) admission, need for ventilation, and death from COVID-19 had been assessed in the literature in association with Gal-3 levels.  Forest plot for the meta-analysis of Gal-3 levels in COVID-19 patients vs. healthy controls after removing the outlier study (33). FIGURE 4 Forest plot for the meta-analysis of Gal-3 levels in severe vs. non-severe COVID-19.

Diagnostic and prognostic use of Gal-3
With

Gal-9 and COVID-19
Five studies investigated the association between Gal-9 and COVID-19. Bai et al. (26) reported significantly higher levels of fulllength Gal-9 in COVID-19 associated with pneumonia than in healthy control (P < 0.0001). In line with these results, two other studies reported higher levels of Gal-9 in SARS-CoV-2-infected individuals compared to non-infected ones (20,28).
Further to these studies, Chen et al. (21) reported that patients with both severe and non-severe forms of COVID- 19  For their part, Bozorgmehr et al. (28) confirmed the results above, showing an increase in Gal-9 in COVID-19 patients compared to healthy controls. In addition, these authors determined a cutoff of Gal-9 plasmatic concentration of 2,042 pg/ ml that allowed the COVID-19 diagnosis with a sensitivity of 94.96% and specificity of 94.92%.
Finally, against this trend, the study from Yaş ar et al. failed to find a statistical difference between mild COVID-19 and healthy controls (19). However, these authors found that severe and critical cases of the disease had significantly higher Gal-9 levels than patients with mild symptoms or healthy controls.

Discussion
Overall, our study demonstrated that the assessment of variation in circulating Gal-1, -3, -8, and -9 levels in response to SARS-CoV-2 infection could be used as potential diagnostic and prognostic non-invasive biomarkers of COVID-19.
In detail, our systematic review revealed that Gal-1 release is increased in COVID-19 patients compared to controls. However, the data about its relevance to disease severity was contradictory. For instance, in one study, it has been reported no significant difference between severe and non-severe COVID-19 patients (22). In comparison, another study reported a higher level of Gal-1 in critical cases compared to those with severe or mild COVID-19 (35).
Concerning Gal-3, our meta-analysis showed that the circulating levels of Gal-3 were significantly higher in COVID-19 patients than in healthy controls, with no significance between severe and non-severe cases. Moreover, our review support that Gal-3 levels are positively associated with COVID-19 complications. A huge body of literature indicates cytokine release as the main contributing factor to morbidity and mortality of COVID-19 (39). Gal-3 is also correlated with several inflammatory biomarkers, which may be the pathophysiology behind this capability to be a biomarker. Classic inflammatory biomarkers such as C-Reactive Protein (CRP), Ferritin, neutrophil count, and D-dimer have been shown to have associations with disease severity (40,41); however, as stated by Cervantes-Alvarez et al., there is collinearity between them. In this study, after multivariate analysis, only CRP and Gal-3 were recognized as independent predictors of disease severity (31). Moreover, due to various cut-offs available for CRP, Gal-3 may even be a better candidate for measurement to stratify disease severity. However, it is worth noting that three reports are currently contrasting with this trend. For instance, Bruni et al. (29) (33) reported that Gal-3 levels are lower in severe COVID-19 patients than in healthy controls.
In addition, despite limited, our systematic review of Gal-8 and -9 showed that lower Gal-8 and higher Gal-9 levels could be observed in COVID-19 patients compared with healthy controls. Concerning the correlation between Gal-8 and -9 to the disease severity, the data available are not conclusive. Indeed, our study showed a tendency toward higher levels in severe cases of Gal-9.
Conversely, for Gal-8 the data are very limited as only one study investigated whether the serum level of this factor is differently modulated by COVID-19 severities (19). In this context, Yasar and colleagues reported a reverse trend of Gal-8 levels compared with other galectins. Nevertheless, it is worth noting that recent studies revealed Gal-8 binds exclusively to a specific motif of the viral receptor-binding domain (RBD) of SARS-CoV-2, called 30SLacNAc, while other galectins identify and bind to other additional motifs of the RBD as well (42). However, more studies are required to confirm this different trend seen in Gal-8 and look for possible explanations as well.
Together, these data suggest that further investigation into the clinical relevance and biological role of Gal-8 and -9 is needed. Indeed, understanding the role of these galectins will not only increase their diagnostic and prognostic significance but may also have therapeutic use in the fight against COVID-19 as well as other inflammatory disorders, as per Gal-1 and -3 which are currently considered two of the most critical therapeutic and drug target in various diseases (43). Studies regarding their use in COVID-19 patients showed that inhibition of Gal-3 could modulate the host immune response and reduce cytokine storm syndrome, a major cause of death in COVID-19, alongside preventing viral entry and reducing post-COVID-19 pulmonary fibrosis (44, 45). Also, different clinical trials are currently targeting Gal-3 for treating fibrosis in various diseases, such as idiopathic pulmonary fibrosis and non-alcoholic steatohepatitis-related liver fibrosis (13). In this context, a phase Ib/IIa randomized controlled platform trial is evaluating the potential effects of the Gal-3 inhibitor GB0139 in hospitalized patients with confirmed COVID-19 pneumonitis (46).

Limitations
There are some study limitations to note. First, Gal-1, -8, and -9 lacked sufficient original studies to perform a meta-analysis. Secondly, regarding the measurement methods, the studies used various techniques and equipment that consequently resulted in heterogeneity, which despite performing meta-regression, may add limitations. Finally, with the present data, we could not compare the effectiveness of galectins as diagnostic and prognostic tools to each other. Thus, additional studies are needed to compare different galectins not only with each other but also with other known COVID-19 biomarkers.

Conclusion
With the emergence of the COVID-19 pandemic, the healthcare systems now face a monumental challenge worldwide to find effective treatments and develop tools for disease diagnosis and monitoring disease progression. To our knowledge, this is the first comprehensive systematic review and meta-analysis evaluating the role of Gal-1, -3, -8, and -9 as potential serum and/or plasma biomarkers for the diagnosis and prognosis of COVID-19.
Although we have identified several galectins differentially modulated in response to SARS-CoV-2 infection, our analysis supports a potential prognostic and diagnostic role for Gal-3, in particular, due to its elevated levels in patients compared with healthy controls and its positive correlation with the disease severity. However, additional studies are needed to corroborate our findings and further examine Gal-3's levels in conjunction with other diagnostics or clinical parameters assessing host immunity, infection history, and other complications associated with SARS-CoV-2 infections. Similarly, further investigation is necessary to shed light on the role of the Gal-1, -8, and -9 that, based on the evidence provided here, may have a similar valuable diagnostic and prognostic role and be at the same time considered a potential therapeutic target, but that remain still poorly studied.