Characterization of IgG N-glycan patterns in COVID-19, sepsis and healthy subjects Authors: Victoria Paredes-Orejudo, Yosra Helali, Axelle Bourez, Alexandre Rousseau, Pierre Van Antwerpen, Karim Zouaoui Boudjeltia, Michael Piagnerelli, Arnaud Marchant, Cédric Delporte Published: 2026 ABSTRACT AbstractPatients with severe infectious diseases may require admission to intensive care units (ICUs) when the infection progresses to severe forms, as observed in sepsis or, more recently, in COVID-19. Despite advances in critical care, the risk of progression toward severe or fatal outcomes remains difficult to predict, highlighting the need for reliable prognostic biomarkers. Immunoglobulin G (IgG) N-glycosylation has emerged as a key modulator of immune responses and may represent a valuable biomarker for assessing disease severity. In this study, we investigated the IgG N-glycans profiles to evaluate their potential association with clinical outcome in critically ill patients with sepsis or COVID-19. Serum IgG N-glycans were characterized using online hydrophilic interaction liquid chromatography (HILIC) solid-phase extraction, followed by HILIC separation and fluorescence and mass spectrometry detection. The study cohort included healthy controls, ICU patients with sepsis or severe COVID-19, the latter being further stratified according to survival outcome. Distinct IgG N-glycosylation patterns and relative abundances of specific N-glycan structures in serum were associated with disease etiology (sepsis versus COVID-19) as well as with patient outcomes for COVID-19 group. These findings suggest that IgG glycosylation profiling may provide a promising approach for stratifying ICU patients according to their risk of developing severe or fatal diseases.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1038/s41598-026-57666-0.Subject terms: Biomarkers, Diseases, Immunology, Medical research, Microbiology Introduction Sepsis is a clinical condition that is caused by inappropriate host response to infection and can lead to septic shock with multiple organ failure resulting in death1. Infection may be due to different pathogens such as bacteria, viruses or fungi. The immune system’s role in this affection is fundamental. Indeed, by detecting a pathogen, different pathways are activated in the immune cells and causes cytokine production and inflammation. The immune system tends to have an exaggerated response to blood stream infections. It is responsible for the aggravation of the general state of the patient and can result in death2. Sepsis remains the leading cause of mortality worldwide in intensive care units (ICUs) with 30–50% mortality rate3. Sepsis mortality is driven by bidirectional immune dysregulation characterized by initial excessive inflammation (IL-6 > 1,000 pg/mL). This is followed by profound immunosuppression marked by lymphocyte apoptosis, T cell exhaustion, and monocyte HLA-DR downregulation, which collectively impair antimicrobial defence and drive secondary infections accounting for 44.2% mortality in affected patients. This immune dysregulation operates through identical inflammatory mediators (IL-6, TNF-α, CRP) that independently predict cardiovascular mortality in general populations without active infection, establishing a mechanistic link between inflammation severity and fatal outcomes across both acute sepsis and chronic disease states4–8. Low IgG levels at sepsis onset independently predict increased 28-day and 90-day mortality (odds ratio 4.07–4.5 in septic shock patients), with hypogammaglobulinemia occurring in 24–61% of septic patients and conferring 1.3–1.7-fold increased mortality risk compared to normal IgG levels. While the mechanism involves impaired bacterial opsonization and phagocytosis, as IgG mediates both innate immune recognition through Fc receptors and complement-mediated clearance of pathogens, and IgG N-glycosylation alterations independently predict hospital mortality risk. Intravenous immunoglobulin (IVIG) therapy targeted to patients with low baseline IgG reduces 28-day mortality by 6.2–8.4% and represents a potentially modifiable risk factor, suggesting that IgG deficiency identifies a high-risk sepsis subgroup that is currently overlooked in clinical guidelines and amenable to immune supplementation therapy9–13. Another cause of ICU hospitalizations is due to respiratory manifestations caused by the SARS-CoV-2 which is the causative agent of the global COVID-19 pandemic that emerged in 2019. Since 2020, over 7.11 million individuals worldwide have died of infection by SARS-Cov-214. SARS-CoV-2 infection exhibits marked selectivity in disease severity. Although the majority of infected individuals remain asymptomatic or present with mild, flu-like symptoms (approximately 40–80% depending on detection methods), an unpredictable proportion of patients may develop severe pathology requiring ICU hospitalization due to respiratory failure, and carrying a substantial mortality risk3. Over recent years, numerous researchers have endeavoured to elucidate the mechanisms underlying this disease selectivity, including viral entry mechanisms, immune evasion strategies, and host immune dysregulation; however, current knowledge remains limited, and identification of novel biomarkers—such as inflammatory cytokines, antibody glycosylation patterns, and biochemical markers—is essential to assist clinicians in optimizing the management and treatment of hospitalized patients. Immunoglobulin G (IgG) is the most abundant and versatile antibody in human serum with a concentration of 700 to 1600 mg/dL in healthy adults15. It is a glycoprotein which is produced by an important actor of the adaptative immune system, the plasma cells, a specific type of lymphocyte B. The immunoglobulins group show five different types of glycoproteins: IgA, IgD, IgE, IgG and IgM. IgG has 4 subtypes, numbered according to their concentration in blood. IgG1 has the highest concentration in human serum compared to the others. These proteins play an important role in the immune response during an infectious disease such as COVID-19 or sepsis16. Post translational modifications are inherently implicated in the activities of any given protein. Amongst them, the glycosylation is a fundamental part of the activity of IgG. The latter has a constant N-glycosylation site on the IgG Fc chain, more specifically, on the asparagine in the sequence type: asparagine-X-threonine/serine17. Furthermore, 10–14% of human serum IgG are also glycosylated on Fab part18. N-glycans have a conserved glycan structure composed by two N-acetylglucosamine (GlcNAc) and three mannoses (Man). Galactose (Gal), fucose (Fuc), sialic acid and GlcNAc can also complete the glycan chain to form the more or less complex N-glycans (Fig. 1). High-mannose N-glycan are predominantly found on therapeutic antibodies and are less frequent in healthy patients. However, they have been observed in some cases of inflammation and cancer, making its investigation interesting19. Therefore, those N-glycans are characterised using those glycan chains16. Representation of two different types of N-glycans present on human IgG: complex (above) and high-mannose (below) N-glycans. For complex N-glycans, they are further distinguished by the presence of core fucose (Fucosylation), galactose (Galactosylation), N-acetylneuraminic acid (Sialylation) and/or bisecting (Bisecting). The Oxford notation nomenclature was used thorough this work while the Consortium for Functional Glycomics (CFG) was applied to illustrate the N-glycans. The primary objective of this study was to investigate serum IgG N-glycosylation in sepsis and COVID-19 cases and compare those groups. We examined whether IgG N-glycosylation patterns differentiate healthy from infected individuals who were hospitalized in ICU due to their infection. More precisely, we compared IgG-glycosylation profiles between COVID-19 patients with fatal and non-fatal outcomes at ICU admission. Furthermore, we characterized IgG-N-glycosylation variations across both sepsis and COVID-19 patients to determine whether these inflammatory conditions share conserved glycosylation signatures or possess disease-specific glyco-patterns. For this purpose, IgG N-glycans isolated from sera of ICU patient’s (71) and healthy volunteer’s (15) serum sample were analysed using hydrophilic interaction liquid chromatography fluorescence detector mass spectrometry (HILIC-FLD-MS). Results As serum IgG glycosylation role is still controversial in immune response in patient suffering of infection requiring ICU hospitalization, IgG were isolated from serum. Their total N-glycans were released from the protein backbone and patterns were analysed as previously stated20. N-glycans were grouped following their traits as total galactosylation, total sialylation, total fucosylation, total bisecting or high-mannose patterns. The patterns were examined as percentage of total IgG N-glycosylations in each of the four (4) studied groups, namely serum samples from healthy individuals (= Healthy group; n = 15), from patients diagnosed with sepsis (= Sepsis group; n = 19), 25 from individuals who recovered from COVID-19 (= COVID-Recov group; n = 25) and 27 from patients who had a fatal outcome from COVID-19 (= COVID-Fatal group; n = 27). First, the galactosylation trait encompasses the mono and di-galactosylation, which is the presence of one or two galactoses at the end of the glycan core structure. The healthy and COVID-Recov groups showed higher percentage of total galacosylated IgG N-glycans compared to the other groups, with a larger decrease for sepsis group (Fig. 2A). Compared to the healthy group, the COVID-Fatal group appears to show a decrease. This trend was already observed prior to age correction, suggesting that it is unlikely to be the result of over-correction for age. Furthermore, total galactosylation is the only total glycosylation trait which discriminated COVID-Fatal and COVID-Recov groups. Relative abundance of total glycosylation trait of serum IgG (A) Galactosylation, (B) Sialylation, (C) High-Mannose, (D) Bisecting, (E) Fucosylation and (F) Afucosylation in different studied groups : Covid-Fatal (n = 27), COVID-Recov (n = 25), Sepsis (n = 19) and Healthy (n = 15). Kruskal–Wallis test, *p-value < 0.05; **p-value < 0.005; ***p-value < 0.001. The data were corrected for age using linear regression model. The sialylation trait corresponds to the presence of sialic acid also called N-acetylneuraminic acid (Neu5Ac) at the end of the glycan core structure. Mono- or disialylated were majorly observed in the current study. Our results show an increase of sialylation in the COVID-19 groups and a decrease in the sepsis group compared to the healthy group (Fig. 2B) with di-sialylated N-glycans leading the trend of increase (Figure S1). Age does not seem to impact those results as we observed the same trends before the age-correction. The high-mannose trait corresponds to the addition of mannose residues to the N-glycan core structure. Our results showed that, in COVID-19 groups, there was a slight increase in high-mannose N-glycans compared to the sepsis and healthy groups even if not statistically different for the healthy group compared to COVID-19 groups when data were corrected for age (Fig. 2C). Furthermore, Fig. 3 shows the two high-mannose N-glycans that were detected, namely M5 and M6 which are responsible for increased high-mannose presence in COVID groups. Relative abundance of M5 and M6 high-mannose N-glycans in COVID-Fatal (n = 27), COVID-Recov (n = 25), Sepsis (n = 19) and Healthy (n = 15) groups. Kruskal–Wallis test, *p-value < 0.05; **p-value < 0.005; ***p-value < 0.001. The data were corrected for age using linear regression model. The bisecting trait is the addition of a GlcNAc to the central mannose in the core structure of the N-glycans. No statistical differences were observed between groups for total bisecting (Fig. 2D). The fucosylation trait is characterised by the presence of a fucose residue on the first N-acetylglucosamine (in the core structure). Both COVID groups (Recov and Fatal) show a decrease while an increase appears in the sepsis group compared to the healthy group (Fig. 2E). A similar but opposite tendency was observed for afucosylation (Fig. 2F). Overall, in the COVID groups, we observed a decrease in total fucosylation and an increase in total sialylation compared to the healthy group which is opposite to observations in the sepsis group. A slight increase in total high mannose was also observed. The higher sialylation seems to be caused by the increase di-sialylation (A2G2S2 & FA2G2S2, Figure S1). For the sepsis group, we observed a decrease in total galactosylation and total sialylation. A slight increase was also found in total fucosylation while the high mannose N-glycans were quite absent. For fucosylation, a decrease in the COVID group and increase in the sepsis group seems to highlight the different IgG glycosylations involved in those two pathologies. Discussion Our results showed several differences of IgG total N-glycosylation between groups. As age is known to have impact on IgG glycosylation, the data were corrected for age effect21,22. This was of particular importance as the healthy group was significantly younger than the other groups. A decrease in galactosylation and sialylation was observed in the sepsis group compared to the other groups. The decrease could be explained by the variation of glycosylation over age as the mean age of each group is quite different (see Material and method section). Although inflammation is also present in COVID-1923, a slight decrease in galactosylation is observed in our COVID-Fatal group compared to the healthy group when the age correction was applied (without age correction, galactosylation was higher). For the COVID-Recov group, the total galactosylation is not statistically different compared to the healthy group. This result is different from most recent studies about galactosylation in patients suffering from COVID-19. Indeed, they show lower galactosylation and, therefore, sialylation in severe cases compared to mild cases16,24–26. Vincente et al.16 investigated the importance of galactosylation loss in COVID by analyzing the agalactosylated IgG FcγRIIIa binding. The FcγRIIIa is a receptor found on natural killer (NK) cells which binds IgG-complexes and induce the production of pro-inflammatory cytokines and cytolytic mediators such as perforins. They are, therefore, an actor of antibody-dependant cellular cytotoxicity (ADCC)27. They found agalactosylation caused an increased binding activity and, therefore, an increased NK cell activation and activity. This galactose loss is correlated to a pro-inflammatory state associated with higher interferon-γ and tumor necrosis factor-α. That phenomenon was also associated with a grimmer outcome of the concerned disease. To our knowledge, this study is the first to compare serum IgG glycosylation patterns in patients with sepsis and COVID-19. The observed galactosylation modifications between the two groups may underlie the common pathological complications associated with fatal COVID-19 and sepsis. De Haan et al. have not observed a significant difference in galactosylation between the patients with severe or non-severe meningococcal infection. On the other hand, they observed a lower sialylation per galactose for patients with a severe outcome28. The galactosylation is usually decreased during inflammatory pathologies such as sepsis and COVID-1912. Sialylation is a glycosylation linked to galactosylation. Indeed, a galactose substitute is required to the N-glycans sialic acid addition. For both sepsis and COVID groups, correlation between galactosylation and sialylation is consistent and could be explained by sialidase and galactosidase activity brought from infection28,29. The sialylation decrease is associated with an upregulation of ADCC. Indeed, sialic acid absence allows IgG FcγRIIIa binding and, therefore, the activation of different immune cells such as NK, macrophages and neutrophils. The activation consequently induces a pro-inflammatory factor production. It has also been found that asialylated glycans indirectly activated the lectin-initiated alternative complement pathway by reducing the binding affinity to complement-dependant cytotoxicity (CDC). The important decrease in sialylation in the sepsis group may be caused by an increase in extracellular sialidase and a higher clearance of sialylated IgG30,31. The COVID-19 groups show an increase in sialylation. The sialylation trait is linked to anti-inflammatory properties32. That is counterintuitive as COVID-19 shows an important inflammatory component which cannot be explained by lower sialylation of IgG. In this study, we observed an increase in high-mannose N-glycans in the COVID-19 groups, especially M5 and M6 (Fig. 3). High-mannose glycans were almost non-present in the sepsis and even in the healthy group where the age correction brought a higher level of high-mannose glycosylation in this group. It has been found that those “high-mannose” IgG have an increased affinity toward FcγRIIIa and show an ADCC activity. It is comparable to afucosylated IgG. This activity could be caused by the fact that the high-mannose glycans do not present a fucosylation and, therefore, the increase is caused by the absence of the fucose rather than the mannose residues. Those glycans show poor activation of the complement dependent cytotoxicity. The high-mannose N-glycans IgG have a higher clearance compared to the others N-glycans IgG19,33,34. Our work shows a significant decrease in fucosylation in the two COVID groups compared to the sepsis and healthy group. The sepsis group and healthy group differences are not significant even if the sepsis group seems to show a slight increase in fucosylation. Fucosylation status of glycoproteins is due to modifications of proteins inside the cells like FUT8, a fucosyltransferase present in Golgi apparatus. Fucosidases are also majorly acting within cells. They also have been described in the circulation, like plasma alpha-L-fucosidase (FUCA2)35–37. However, the major fucosylation modifications remain an inside cell process. Delbo Larsen et al. sought to explore the implication of a fucosylation decrease in viral infections such as COVID38. The afucosylation (loss of the fucose residue) is implicated in inflammatory response by inducing pro-inflammatory cytokines production such as interleukin-6. The afucosylated IgG shows an increased affinity toward FcγRIIIa and an increased ADCC activity. FcγRIIIa receptor induces proinflammatory cytokine production39. Afucosylated IgG are typically targeted toward enveloped viruses such as COVID. The IgG fucosylation seems to be correlated with the severity of the infection. On the other hand, higher fucosylation, like in the sepsis group, could be the result of an increased level of α1–6 fucosyltransferase. Pro-inflammatory cytokines regulate this enzyme expression38,40,41. Higher fucosylation in sepsis has been observed in experimental murine sepsis by Yaykasli Kursat et al.42. Bisecting N-glycans increase the IgG FcγRIII binding which activate ADCC and inflammation. A peculiar consequence of the addition of a bisecting N-acetylglucosamine is the competitive inhibition of FUT8 by MGAT3. FUT8 is the enzyme responsible to the addition of a fucose to the N-glycan core structure. MGAT3 is an enzyme responsible of the bisecting N-acetylglucosamine substitute in N-glycan. Therefore, bisecting IgG indirectly lowers the IgG total fucosylation with all the implications discussed previously. The bisecting N-glycans FA2B has been reported as of particular interest and increased in inflammation43,44. In our study, when corrected for age, there was no difference observed between the groups (Figure S2). However, FA2B was lower in the healthy group than in the other groups when no correction for age effect was applied (Figure S2B). This could be explained by the age as it was observed that bisecting increase with age33. This tends to comfort that young people are more likely to be protected against severe sepsis and COVID-19 than old ones. Inflammation plays a major role in the severity of both sepsis and COVID-19, as each condition is characterized by a heightened inflammatory response that can become life-threatening. The findings in our sepsis cohort appear somewhat counterintuitive. Indeed, the inflammation seems to not be completely controlled by IgG N-glycans. Otherwise, we should observe reduced galactosylation and reduced fucosylation12. That is not the case as we observe a reduced galactosylation and a higher fucosylation compared to the healthy group. This suggests that IgG glycosylation contributes to the inflammatory state in sepsis but is not the dominant driver. In COVID-19, we observe a reduced fucosylation whereas the galactosylation remains unchanged compared to the healthy group. This indicates a slight proinflammatory action from the IgG N-glycans but, our results show that it is not the main actor in inflammation. In conclusion, we observed a decrease in galactosylation in COVID-19 and sepsis patients with a more pronounced decrease for COVID-Fatal and sepsis groups. This decrease of galactosylation of IgG induced a pro inflammatory environment because of its higher binding to FcγRIIIa on NK cells, that could favour the exaggerated immune response. As this decrease is more important in COVID-Fatal and sepsis groups, it could be an interesting indicator of severity. The sialylation increase in the COVID group is counterintuitive and could be caused by the presence of disialylated glycans. The sepsis group show a decrease in sialylation which is coherent with the sepsis group’s decrease in galactosylation. Asialylated glycans also allow a higher FcγRIIIa binding and, therefore, an ADCC upregulation. High-mannose glycans were increased in the COVID group compared to the sepsis and control group. Those glycans also show an increased FcγRIIIa binding properties and, in consequence, a pro inflammatory activity. The glycan fucosylation increased in the sepsis group while it decreased for the COVID group. A particular feat is highlighted by that phenomenon. Indeed, even though sepsis and COVID show inflammatory properties, the glycosylation is different depending on the disease. Patients with fatal COVID show a high level of afucosylated glycans. Those glycans promote a higher binding to the FcγRIIIa receptor and pro inflammatory cytokine production. The bisecting glycans increased in the COVID and sepsis patients compared to the control group. Our study has several limitations that may affect the results, notably the small cohort size and the ages differences. Indeed, the largest group, which is the COVID-deceased, contain the samples of 27 patients. The smallest group consist of 15 patients. This may have reduced the statistical power and the generalizability. The information outlined in this study may not represent the general population. The age variation also impacts the analysis. Older patients show variables comorbidities and complex treatments. The proportions of different N-glycan traits such as galactosylation varies with age45. This is why we age corrected our data. Glycosylation studies show different results between one another. Indeed, glycosylation is influenced by external factors such as physical activity, diet, age, sex and genetic factors33. Demographic factors also play into the equation. IgG glycosylation varies thorough the inflammation and disease progression as observed in our results. However, further studies are needed to elucidate glycan heterogeneity and the clinical importance of IgG specific glycans that could be used as biomarkers of severity and even mortality risks on large cohorts of patients in ICU. Materials and methodsSample collection The study cohort included 15 serum samples from healthy individuals, 19 from patients diagnosed with sepsis, 25 from individuals who recovered from COVID-19 and 27 from patients who succumbed from COVID-19. For all the groups beside the sepsis group, the samples were collected during the patient’s first admission day in the intensive care unit in CHU Charleroi-Chimay. The sample collection for the sepsis group was done on between the first and fourth day of hospitalisation in ICU. The age medians in each group were as follows: 70 [65–76] years old for COVID-Fatal group; was 61 [49–79] years old for COVID-Recov group; 30 [23–45] years old for healthy group; and 62 [37–71] for sepsis group. The ethical protocols for this study were submitted to the ethics committees of CHU Charleroi-Chimay (Humani, EC ISPPC OM008, Charleroi, Belgium: P20/32_06/04, CCB: B3252020 and B3252021000003) and Erasme-ULB hospital (2018/129, CCB: B406201835230; 2018/637, NCT03849677, Brussels, Belgium). The informed consent was obtained from all participants, and the experiments were performed in accordance with the relevant guidelines and regulations. Chemicals and reagents The Thermofischer Life Technologies’s Melon Gel IgG Spin Purification kit was used to isolate the samples’ IgG. The samples were digested using Sigma Aldrich sodium dodecyl sulfate (SDS) 1.48% (m/v), Roche Diagnostics Nonidet P40 4% (m/v), Peptide N-glycoside F from Promega and a phosphate buffered saline (PBS) from VWR. The labelling was done with Merck Procainamide hydrochloride (ProcA), 2-methylpyridine borane complex (2-PB) from Sigma Aldrich and a solution of dimethylsulfoxide/acetic acid (70/30, v/v) from, respectively, Sigma Aldrich and Fisher Chemical. N, N-Dimethylformamide (DMF) from VWR and acetonitrile (ACN) from Fisher Chemical were used to stop the digestion. The HILIC-FLD-MS was carried on with Agilent 6520 A Accurate-Mass Quadrupole Time-of-Flight (QTOF) with dual electrospray (ESI). The solvents used during the enrichment and separation of the IgG were water (Fisher Chemical), ammonium formate 99% (VWR) and ACN (Fisher Chemical). The analysis was carried out in MS1. Sample preparation The sample was purified to only contain IgG. Then, the digestion allows the glycans liberation. The final step is to label those glycans with procainamide. The serum purification was realised with Melon Gel IgG spin purification kit from Thermofischer Life Technologies. Briefly, the column was filled with Melon Gel and centrifugated. The gel matrix was subsequently washed twice to remove residual contaminants. In parallel, the serum was diluted and then loaded onto the gel and incubated with end-over-end mixing to facilitate binding. The purified sample was finally recovered after the centrifugation of the column. This sample preparation protocol was previously developed and validated by Helali Y. et al.20. The samples were then digested and labelled using the long protocol developed and validated by Helali Y. et al.20. Briefly, extracted IgG were denaturated with SDS, then digested using PNGase F. Afterward, the labelling solution, composed of ProcA and 2-PB, is added and incubated. ProcA is used as a fluorescent tag while the 2-PB functions as a reducing agent to stabilise the final adduct formed between ProcA and the N-glycan. The reaction is stopped using DMF and ACN. Online HILIC SPE HILIC-FLD-MS analysis The system used consists of an online HILIC SPE system coupled with a HILIC-FLD-MS. During the online SPE, the column used was a XBridge BEH Amide 2.5 μm, 2.1 × 30 mm XP from Waters. An Agilent quaternary pump serie 1100 was used to enrich the previous column with the sample’s glycans. Then, the glycans are flushed toward the analytical column. It is a XBridge BEH Amide 2.5 μm×150 mm XP column from Waters. Then, the glycans are separated, with part of it passing through a FLD detector and the other part through a mass spectrometer. The mobile phase A was a 50 mM ammonium formate adjusted to pH 4.4 while the mobile phase B consisted in pure acetonitrile. For the enrichment part of the analysis, the gradient started at 80% of ACN with a flow rate of 0.6 mL/min for a minute. The flow rate stays the same for the whole enrichment part. Then, at 1.1 min, the mobile phase B is lowered to 77% of ACN. At 4 min, the mobile phase B is lowered to 10%. At 19 min, the mobile phase B is at 80% and stays the same until the end at 33 min. For the separation on the analytical HILIC column, the starting gradient was 77% ACN with a flow rate of 0.5 mL/min. At 7 min, the mobile phase B was 68% with the same flow rate. At 31 min, the mobile phase B was 57% with the same flow rate. At 32 min, the mobile phase B was 20% and the flow rate, 0.25 mL/min. At 37.5 min, the gradient was 77% of ACN with a flow rate of 0.25 mL/min. At 42 min, the mobile phase B was 77% with a flow rate of 0.5 mL/min until the end of the analysis ate 46 min. The analysis was done on the Quadrupole-Time-Of-Flight (QTOF) series 6520 A from Agilent Technologies following the method developed and validated by Helali Y. et al.20. Quantification was performed using MS signal intensities following the approach described by Helali Y. et al.20. The calculation of total glycosylation trait and the N-glycan peak information are presented in supplementary information. Statistical tests The statistical analyses were performed with SigmaPlot®12 (Systat Software INC) by using Kruskal–Wallis One Way analysis of Variance on Ranks test with all pairwise multiple comparison procedure (Dunn’s Method). The data was corrected for age using linear regression model. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Author contributions The samples were collected by AM, KZB and MP. The experimental design was done by CD, YH, PVA, MP, KZB and AM. The data analysis and statistical analysis were done by YH, VP-O, PVA, MP, AM, KZB, AB and CD. YH prepared the samples and carried out analysis and AB helped performing QTOF and glycan analysis. VP-O, YH and CD wrote the first draft and carried out the final glycan analysis. CD supervised the project. All authors have given approval to the final version of the manuscript. The manuscript was written and verified through contributions of all authors. Funding This project has been funded by Belgian National Fund for Scientific Research (FRS-FNRS) (grants FRS N° 3.4553.08, J 0.0052.20 CDR and J.0061.22 CDR) and Université libre de Bruxelles (FER-2007, 2019 & 2023). Data availability The datasets generated and analysed during the current study are available from the corresponding author on reasonable request. Declarations Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Victoria Paredes-Orejudo and Yosra Helali contributed equally to this work. References Associated Data Supplementary Materials Supplementary Material 1 Data Availability Statement The datasets generated and analysed during the current study are available from the corresponding author on reasonable request.