Kawasaki Disease (KD) is the leading cause of acquired heart disease in children. Despite its growing prevalence, the exact etiology and underlying biological mechanisms of KD remain largely unresolved. As a consequence, there is still no diagnostic test for KD. To fill this critical gap, we examined circulating cell-free RNA (cfRNA) in plasma as a biomarker to diagnose KD.
Plasma cfRNA offers broad clinical utility because it represents a mixture of transcripts from all vascularized tissues. cfRNA can measure tissue-specific changes in gene expression, intra- and intercellular signaling, and cell death, making it ideal for identifying disease processes in different tissues throughout the body. Here, we profiled the plasma cfRNA transcriptome of 345 KD patients and 340 febrile controls who presented with at least one of the clinical criteria for KD.
Machine learning using a GLM with Lasso regression identified a gene signature for KD which achieved a test AUC 0.93 and an accuracy of 87%. Notably, this model only used transcripts upregulated in KD and accurately identified patients against a heterogenous background of over 40 unique febrile conditions, including scarlet fever. The youngest patients (<1 year) had poorer performance relative to older KD cases. We also found that the performance of this cfRNA signature was superior in patients with higher CRP and those with more severe coronary artery aneurysms. Cell-type deconvolution of the cfRNA transcriptome identified a large increase in neutrophil and hematopoietic stem cell transcripts in KD patients. Pathway analysis identified elevated neutrophil degranulation in KD and increased interferon and cytokine signaling in febrile controls. Lastly, we noted significant differences in cfRNA transcripts and molecular pathways between clinical subgroups of KD.
This study is ongoing. We plan to repeat this analysis and improve our current diagnostic model once we have sequenced 1,000 samples. Future analysis will evaluate if cfRNA has the potential to serve as a surrogate biomarker for a clinical trial or if it can detect children at high risk of coronary artery aneurysms.