A Multi-Cohort Study of Immunoglobulin G Glycans in Newly Diagnosed Inflammatory Bowel Disease Patients Reveals Accelerated Biological Aging
Flevaris, K.; Trbojevic-Akmacic, I.; Goh, D.; Lalli, J. S.; Vuckovic, F.; Capin Vilaj, M.; Stambuk, J.; Kristic, J.; Mijakovac, A.; Ventham, N.; Kalla, R.; Latiano, A.; Manetti, N.; Li, D.; McGovern, D. P. B.; Kennedy, N. A.; Annese, V.; Lauc, G.; Satsangi, J.; Kontoravdi, C.
Show abstract
Background and AimsAlterations in immunoglobulin G (IgG) N-glycosylation are implicated in inflammatory bowel disease (IBD); however, the robustness of IgG glycan signatures across IBD cohorts with diverse demographics and geographic origins remains underexplored. We aimed to determine whether compositional data analysis (CoDA) and machine learning (ML) can identify IBD-related IgG N-glycan signatures and whether these signatures capture disease-associated acceleration of biological aging. MethodsWe analyzed the IgG glycome profiles of 1,367 plasma samples collected from healthy controls (HC), symptomatic controls (SC), and people with newly diagnosed Crohns (CD), and ulcerative colitis (UC) across four cohorts (UK, Italy, United States, and Netherlands). IgG glycosylation was analyzed by ultra-high-performance liquid chromatography, yielding 24 total-area-normalized glycan peaks (GPs). Analyses were performed using cross-sectional data obtained at baseline. CoDA-powered association analyses were used to identify disease-related effects on GPs while controlling for demographic covariates. ML models were trained and evaluated to assess generalizability to unseen cohorts and demographic subgroups, with a focus on discrimination and reliability. ResultsAcross all cohorts, people with IBD demonstrated accelerated biological aging as quantified by the GlycanAge index. This was accompanied by consistent reductions in IgG galactosylation, with effects partially modulated by age. Classification models trained on glycomics and demographics achieved robust discrimination (AUROC{approx}0.80) between non-IBD (HC+SC) and IBD across cohorts. ConclusionThese findings reveal accelerated biological aging in people with IBD and support the translational potential of IgG glycans as biomarkers and a novel route toward clinically interpretable personalized risk estimates.
Matching journals
The top 7 journals account for 50% of the predicted probability mass.
Similar papers in this journal
Similar papers in this journal
- Antigen-driven expansion of public clonal T cell populations in inflammatory bowel diseases 92%
- Intestinal receptor of SARS-CoV-2 in inflamed IBD tissue is downregulated by HNF4A in ileum and upregulated by interferon regulating factors in colon 92%
- Mapping adaptive immune responses toward fungal antigens in inflammatory bowel disease using T cell repertoire sequencing and phage-immunoprecipitation sequencing 92%
Similar papers in this journal
- Deep Proteome Profiling of Metabolic Dysfunction-Associated Steatotic Liver Disease 93%
- The Interpretable Multimodal Machine Learning (IMML) framework reveals pathological signatures of distal sensorimotor polyneuropathy 92%
- Blood and tissue correlates of steroid non-response in checkpoint inhibition-induced immune-related adverse events 92%
Similar papers in this journal
- Assessment of infant outgrowth of cow's milk allergy in relation to the faecal microbiome and metaproteome 92%
- Interpretable machine learning identifies paediatric Systemic Lupus Erythematosus subtypes based on gene expression data 91%
- Globally elevating the AGE clearance receptor, OST48, does not protect against the development of diabetic kidney disease, despite improving insulin secretion. 91%
Similar papers in this journal
- Highly multiplexed spatial analysis identifies tissue-resident memory T cells as drivers of ulcerative and immune checkpoint inhibitor induced colitis 92%
- Immunoglobulin G glycome composition in transition from pre-menopause to menopause 92%
- Mortality in hemodialysis: Synchrony of biomarker variability indicates a critical transition 91%
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.