0077: Transformer-based multi-omics study identifies important role of glycine, serine and threonine metabolism pathway in rheumatoid arthritis complicated by anemia

Sunday, October 26, 2025

10:30 AM - 12:30 PM Central Time

Location: Hall F1

Abstract Poster Presenter(s)

FZ

Fanxin Zeng, PhD

Departmant of Clinical Research Center, Sichuan Clinical Research Center for Medical Imaging, Dazhou Central Hospital
Dazhou, Sichuan, China

Disclosure(s): No financial relationships with ineligible companies to disclose

Background/Purpose: Anemia is a prevalent hematologic complication of rheumatoid arthritis (RA) that exacerbates the disease process and severely impacts clinical performance and treatment strategies. We propose to explore the potential molecular changes in RA complicated by anemia (RA_ane) based on transformer architecture and comprehensive multi-omics analysis.

Methods: This study recruited 257 RA patients, 113 controls, and disease comparison groups including 88 systemic lupus erythematosus (SLE) and 37 gout patients. Their plasma samples were collected for metabolomics sequencing, fecal samples for 16S rRNA sequencing, and peripheral blood mononuclear cells were used for transcriptomics, proteomics and phosphoproteomics studies. The transformer architecture and difference analysis were used to obtain overlapping difference features, and 5 machine learning algorithms ( k-nearest neighbor algorithm [KNN], support vector machine [SVM], random forest [RF], recursive feature elimination-RF [RFE-RF], and eXtreme gradient boosting algorithm [XGBoost]) were further used to perform significant feature screening and construct clinical diagnostic models.

Results: By integrating the results of the above characterization screen, L-tryptophan, glyceric acid, L-malic acid, ALAS2, CA1, SELENBP1, and RPIA were identified as key biomarkers of RA_ane. Further functional enrichment analyses revealed the core role of the glycine, serine and threonine metabolism (GSTM) pathway in RA_ane and were validated by functional enrichment analyses in other dimensions of multi-omics, in addition to metabolomics and microbiomics analyses of SLE and gout, which indicated that alterations in the GSTM pathway are RA_ane-specific features. Subsequently, we developed and validated multi-omics diagnostic models for RA_ane patients with AUC values ranging from 0.75 to 0.84, and finally, we constructed a molecular map of GSTM-based multidimensional changes in the RA_ane population.

Conclusion: This study demonstrated that GSTM pathway plays a crucial role in RA_ane, shedding light on potential therapeutic interventions.