Identification of biomarkers of response to abatacept in patients with SLE using deconvolution of whole blood transcriptomic data from a phase IIb clinical trial

OBJECTIVE: To characterise patients with active SLE based on pretreatment gene expression-defined peripheral immune cell patterns and identify clusters enriched for potential responders to abatacept treatment. METHODS: This post hoc analysis used baseline peripheral whole blood transcriptomic data from patients in a phase IIb trial of intravenous abatacept (~10 mg/kg/month). Cell-specific genes were used with a published deconvolution algorithm to identify immune cell proportions in patient samples, and unsupervised consensus clustering was generated. Efficacy data were re-analysed. RESULTS: Patient data (n=144: abatacept: n=98; placebo: n=46) were grouped into four main clusters (C) by predominant characteristic cells: C1—neutrophils; C2—cytotoxic T cells, B-cell receptor-ligated B cells, monocytes, IgG memory B cells, activated T helper cells; C3—plasma cells, activated dendritic cells, activated natural killer cells, neutrophils; C4—activated dendritic cells, cytotoxic T cells. C3 had the highest baseline total British Isles Lupus Assessment Group (BILAG) scores, highest antidouble-stranded DNA autoantibody levels and shortest time to flare (TTF), plus trends in favour of response to abatacept over placebo: adjusted mean difference in BILAG score over 1 year, −4.78 (95% CI −12.49 to 2.92); median TTF, 56 vs 6 days; greater normalisation of complement component 3 and 4 levels. Differential improvements with abatacept were not seen in other clusters, except for median TTF in C1 (201 vs 109 days). CONCLUSIONS: Immune cell clustering segmented disease severity and responsiveness to abatacept. Definition of immune response cell types may inform design and interpretation of SLE trials and treatment decisions. TRIAL REGISTRATION NUMBER: NCT00119678; results.