Reproducible Molecular Subtypes in Systemic Lupus Erythematosus Are Associated With Disease Activity, Serology, and Distinct Trajectories Over Time.

OBJECTIVE

Systemic lupus erythematosus (SLE) is biologically heterogeneous, motivating development of reproducible molecular subtypes that can track change over time. The aim of this study was to establish a reproducible molecular stratification of SLE across cohorts and platforms and connect the resulting subtypes to clinical activity, serology, and serum proteomics.

METHODS

We computed Gene Set Variation Analysis scores for eight immune gene modules from baseline transcriptomes (ILLUMINATE whole-blood microarray [N = 1,756], LOTUS whole-blood RNA sequencing [RNA-seq; N = 585], and Genuity peripheral blood mononuclear cell (PBMC) RNA-seq [N = 812]). We clustered ILLUMINATE patients (K = 2-8) and selected K = 6, then used a random forest classifier to assign subtypes in LOTUS and Genuity. We analyzed serum proteomics (SomaScan in LOTUS; Olink in Genuity), assessed glucocorticoid associations with neutrophil modules, and quantified baseline to week 24 stability in LOTUS.

RESULTS

Six molecular subtypes replicated across cohorts and corresponded to different levels of disease activity and serology (P < 0.001). Proteomic patterns converged with transcriptomics, with concordant differential proteins across Olink and SomaScan among shared targets. Neutrophils were a key axis: one neutrophil module tracked glucocorticoid use, whereas a second was glucocorticoid independent. In LOTUS, subtype assignments were mostly stable to week 24; when patients moved, transitions were nonrandom, occurred between adjacent subtypes, and were similar in the placebo group.

CONCLUSION

We demonstrated reproducible SLE molecular subtypes (interferon, plasmablast/B cell, neutrophil) supported by independent proteomics. In LOTUS, subtypes were largely stable through week 24. When patients changed subtypes, transitions followed preferential trajectories, supporting longitudinal tracking and trial design utility.