Related Experiment Videos
Nailfold capillary phenotypes distinguish juvenile myositis subtypes and associate with disease activity
Nicholas McClellan1, Li Chen1, Sophia Matossian1
1Department of Pediatrics, Division of Rheumatology, University of Michigan, Ann Arbor, Michigan, USA.
Background:
Nailfold capillaroscopy is a non-invasive method to visualize altered microcirculation in pediatric rheumatic disease, with potential to aid in diagnostic differentiation and disease monitoring. We used nailfold video capillaroscopy (NVC) and machine learning to identify patterns of capillaroscopic features in juvenile dermatomyositis (JDM) and associations of capillaroscopic features with clinical data.
Methods:
NVC features were quantified using automated neural network-based software in 76 individuals, including 18 controls, 33 JDM, 17 childhood-onset systemic lupus erythematosus (cSLE), and 8 overlap myositis (OM) patients. Unsupervised cluster analysis by capillaroscopic features was performed, and clinical features were characterized by cluster. Differences in capillaroscopic features between disease groups were assessed using the Kruskal-Wallis test. For JDM and OM patients, capillaroscopic and clinical feature associations were assessed using Spearman correlation. In 10 treatment-naïve myositis patients (JDM + OM), changes in capillaroscopic features between diagnosis and 3-month follow-up were evaluated.
Results:
Cluster analysis identified three patient clusters, characterized by the presence of either branched or enlarged capillaries, or absence of these abnormalities. The "branched" cluster was most frequently assigned among JDM and OM and consisted of no controls. The majority of TIF1y+ JDM (4/5) were in the "branched" cluster. In JDM and OM, microhaemorrhage density correlated with disease duration (r = -.45, p = .0033), physician global assessment score (r = .6, p = .0004), lactate dehydrogenase (r = .43, p = .01), von Willebrand factor antigen (r = .51, p = .043), and neopterin (r = .69, p = .01). In 10 myositis patients, microhaemorrhage density decreased while capillary density increased from diagnosis to three months post-treatment.
Conclusions:
A branched capillaroscopic pattern was observed more frequently in JDM and OM, particularly TIF1y+ JDM patients within our cohort. Microhaemorrhage density was the most changeable capillaroscopic feature and associated with markers of increased disease activity.
