The research team, led by Laura E. Eades and Fabien B. Vincent at Monash Health in Melbourne, drew patients from the Australian Lupus Registry and Biobank, deliberately balancing the cohort between people of European and Asian ethnicity and across levels of disease activity. Ninety percent of participants were women, the median age was about 42 years, and the median disease duration exceeded twelve years. Crucially, the study did not stop at a single snapshot: the researchers mined longitudinal clinical records stretching back years, calculating how much cumulative time each patient spent in remission or low disease activity, which organ systems had ever been active, and how much organ damage had accumulated.

The technology behind the scan is worth understanding. The i-Ome Discovery Protein Array, developed by Sengenics, uses a patented protein-folding technology called KREX to present 1,613 full-length, correctly folded human antigens on a single slide. Serum from each patient is incubated on the array, and bound IgG autoantibodies are quantified as relative fluorescence. The array was customised to include three interferons, IFN-alpha 1, IFN-gamma and IFN-omega, reflecting the central role of interferon signalling in lupus pathogenesis. After rigorous normalisation and filtering, 533 autoantibody specificities survived for analysis.

To make sense of this high-dimensional data, the team applied unsupervised clustering, a machine-learning approach that groups patients without any prior assumptions about their clinical features. Using Partitioning Around Medoids with bootstrap stability testing over 500 resamples, four robust patient clusters emerged. One cluster stood out: patients characterised by autoantibodies against La and Ro60, two classic lupus antigens, were significantly more likely to be of Asian ethnicity and had higher erythrocyte sedimentation rates than patients in other clusters. Beyond these associations, however, the broad cluster signatures did not map cleanly onto conventional clinical categories, suggesting that individual autoantibody specificities may carry more clinically useful information than global autoantibody patterns.

The individual autoantibody analysis produced the study’s most intriguing results. Anti-LRRFIP2, an autoantibody against leucine-rich repeat flightless-interacting protein 2, was positively associated with being in DORIS remission at the time of blood sampling, and with spending a greater proportion of follow-up time in both remission and the lupus low disease activity state. In supervised machine learning using LASSO regression, anti-LRRFIP2 was the most consistently selected feature in a model predicting DORIS remission, and that model was the only one among twenty tested that demonstrated predictive performance exceeding chance under repeated hold-out resampling and permutation testing.

LRRFIP2 is a pro-inflammatory protein that activates NF-kappaB signalling by binding to the cytosolic portion of Toll-like receptor 4 and to MyD88, key components of innate immune activation. The paradox that autoantibodies against a pro-inflammatory protein correlate with quiescent disease raises the possibility that anti-LRRFIP2 may exert a regulatory or neutralising effect on its target. Naturally occurring non-pathogenic autoantibodies are known to contribute to immune homeostasis, and regulatory autoantibodies targeting type I interferon have previously been described in lupus and associated with lower disease activity. Supporting this line of thinking, the researchers found in a public RNA sequencing dataset of 1,760 lupus patients that LRRFIP2 gene expression was positively associated with disease activity and with TLR4 expression, consistent with the protein’s pro-inflammatory role and lending biological plausibility to the idea that antibodies against it might dampen inflammatory signalling.

Perhaps the most evocative discovery concerned the brain. Anti-gephyrin autoantibodies were associated with central nervous system lupus in the cohort. Gephyrin is a scaffolding protein that organises inhibitory neurotransmitter receptors at GABAergic synapses, and autoantibodies against it have previously been identified in a patient with Stiff-Person Syndrome, a rare autoimmune neurological disease, and more recently in patients with systemic sclerosis, where they were linked to gastrointestinal dysmotility. The finding that this same specificity appears in lupus patients with CNS involvement suggests a possible shared autoimmune mechanism targeting inhibitory synapses, though the authors caution that only six patients in the cohort had CNS lupus and the corresponding machine-learning model did not outperform chance in robustness testing.

Three other autoantibodies showed a different pattern: anti-AAK1, anti-RPS6KB1 and anti-RPS6KB2 were negatively associated with mucocutaneous lupus, meaning patients with higher levels of these autoantibodies were less likely to have had active skin and mucosal disease. RPS6KB1 and RPS6KB2 are ribosomal protein S6 kinases involved in insulin signalling and cellular inflammation, while AAK1 plays roles in viral infection susceptibility and neuropathic pain. None of these three proteins had previously been described as autoantibody targets in autoimmune connective tissue disease, making the finding entirely novel, though again limited by the small number of patients with mucocutaneous activity in the robustness analysis.

In a targeted analysis of interferon pathway autoantibodies, the team found that anti-IFN-omega was negatively associated with adjusted mean disease activity, organ damage and biopsy-proven lupus nephritis, but positively associated with mucocutaneous involvement. Anti-IRF5, a transcription factor central to type I interferon induction, was positively associated with active renal disease, aligning with prior reports linking IRF5 to lupus severity. Anti-IFN-alpha 1 was associated with musculoskeletal involvement, generally a less severe phenotype. These targeted analyses were exploratory and not corrected for multiple testing, so they should be interpreted as hypothesis-generating rather than definitive.

The authors are careful to frame the study as a biomarker-discovery exercise rather than a validation trial. The cohort of 60 patients, while well characterised and ethnically balanced, is modest, and several clinical subgroups were too small for reliable statistical modelling. Autoantibody profiling was performed at a single time point, so associations with longitudinal outcomes reflect correlation with overall disease phenotypes rather than true prediction of future disease course. Nonetheless, the identification of anti-LRRFIP2 as a candidate biomarker for remission and low disease activity, and anti-gephyrin as a candidate marker for CNS lupus, opens concrete avenues for follow-up. If validated in larger independent cohorts, these autoantibodies could eventually inform prognosis, guide treatment de-escalation, or even point to new therapeutic targets in a disease where the immune system’s own weapons may, in some cases, be working in the patient’s favour.

Subject of Research: High-plex serum IgG autoantibody profiling in systemic lupus erythematosus and its associations with disease activity, remission and organ-specific phenotypes

Article Title: Serum autoantibody profiling in systemic lupus erythematosus reveals associations with low disease activity and remission

Article References: Eades, L. E., Figgett, W. A., Lennard, K., Valipour Kahrood, H., Hoi, A. Y., Kandane-Rathnayake, R., Morand, E. F., Papadaki, A., & Vincent, F. B. (2026). Serum autoantibody profiling in systemic lupus erythematosus reveals associations with low disease activity and remission. Frontiers in Lupus, 4, Article 1892328. https://doi.org/10.3389/flupu.2026.1892328

Image Credits: AI Generated

DOI: 10.3389/flupu.2026.1892328

Keywords: systemic lupus erythematosus, autoantibodies, biomarkers, remission, LRRFIP2, gephyrin, interferon, protein array, machine learning, lupus nephritis, CNS lupus, immunology