Juvenile dermatomyositis (JDM) is a rare and heterogeneous autoimmune and inflammatory myopathy characterized by specific cutaneous and muscular involvement. Despite therapeutic advances, cohort studies suggest that only about half of patients achieve treatment-free remission, and relapses can occur even after several years of clinical stability. The pathophysiology combines strong and selective type I interferon activation in the circulation and tissues, a vasculopathy with capillary loss and muscle ischemia, as well as altered mitochondrial biology. Early identification of severe forms remains an unmet medical need, limited by the still partial understanding of the cellular and molecular mechanisms underlying the variability of disease course.
To explore this heterogeneity, the authors applied multicellular-resolution spatial transcriptomics (Visium technology) to muscle biopsies, coupled with reference-free deconvolution and standardized morphometry. Two JDM biopsies of differing clinical severity at diagnosis were compared with healthy muscle, after which the identified signatures were assessed in two additional biopsies from a single patient, before treatment and after two years of remission. The reference-free approach allows molecular signatures to be isolated from the spatial data alone, without the risk of excluding affected muscle fibers or other cell types.
This analysis revealed the disappearance of the normal muscle signature, corresponding mainly to mitochondrial biology, in favor of three distinct pathological signatures: "myofibrillar stress," "muscle remodeling," and "interferon signaling." The myofibrillar stress signature predominated in the most severe biopsy (acute NXP2+ form) and was associated with ischemic fiber lesions; its most overexpressed gene, ANKRD1, had never previously been linked to JDM. Conversely, the muscle remodeling signature, notably marked by FSTL1, dominated in the patient with a favorable course. The interferon signature was present in all active patients but was distributed very heterogeneously, suggesting non-uniform production or response within the muscle.
Examination of the muscle after two years of remission, achieved under ruxolitinib, showed a marked reduction in all three signatures, but only the interferon signature had normalized, whereas the myofibrillar stress and remodeling signatures persisted. This persistence, despite clinical and histological remission, may shed light on the risk of late relapse. The authors emphasize the limitations of this pilot study — the absence of complementary nuclear sequencing, the small sample size, and its descriptive nature — and call for confirmation of these findings in a larger cohort. This work paves the way for a better understanding of the pathophysiology of affected muscle and for the identification of biomarkers predictive of relapse.