Related Experiment Video
Updated: Jun 25, 2026
High-Throughput Transcriptome Analysis for Investigating Host-Pathogen Interactions
Published on: March 5, 2022
Integrated miRNAome-transcriptome analyses identify an immuno-hematopoietic subcluster in patients with long COVID
Ying Yang1, Mari Kanerva2, Helena Liira3
1Institute of Environmental Medicine, Karolinska Institutet, Stockholm, Sweden; Human Microbiome Research Program, Faculty of Medicine, University of Helsinki, Helsinki, Finland.
Background:
Persistent symptoms following SARS-CoV-2 infection, termed post-COVID-19 condition or long COVID (LC), impose substantial psychological and socioeconomic burdens. However, microRNA (miRNA)-mRNA interactions underlying LC heterogeneity remain incompletely defined.
Objective:
We sought to determine whether integrative blood miRNA-mRNA profiling identifies molecular LC subclusters linked to clinical and immuno-hematopoietic features.
Methods:
We performed integrated miRNAome-transcriptome profiling of circulating blood RNA from individuals with LC and recovered control participants. Differential miRNA and mRNA expression was assessed using limma software, and hierarchical clustering was used to define LC subclusters. Validated miRNA-mRNA interaction networks were constructed using the miRNet tool. Clinical, functional, and biochemical parameters were compared between subclusters, and a random forest classifier was developed.
Results:
Clustering identified an immuno-hematopoietic LC subcluster, LC1, characterized by extensive miRNA-mRNA dysregulation; enrichment of erythropoietic, platelet, and immune pathways; and biochemical alterations including lower plasma sodium and elevated fibrin D-dimer and thrombin time. Compared with subcluster LC2, LC1 showed persistently greater symptom burden, functional impairment, reduced quality of life, lower resilience, and higher anxiety/depressive symptoms. Potential confounding by age, sex, comorbidities, and selected medication use was evaluated. Network and coexpression analyses identified regulatory nodes enriched for viral infection and natural killer cell activation pathways. A random forest classifier incorporating 9 miRNAs and LRRFIP2 achieved an area under the receiver operating characteristic curve of 0.91 for LC1, distinguishing LC1 from LC2 and recovered control participants.
Conclusions:
LC comprises biologically and clinically distinct subclusters shaped by coordinated miRNA-mRNA remodeling. The immuno-hematopoietic LC1 subcluster supports biomarker-based stratification of patients with persistent physiologic and clinical impairment.
