BackgroundMonths after infection with SARS-CoV-2, at least 10% of patients still experience complaints. Long-COVID is a heterogeneous disease and clustering efforts revealed multiple phenotypes on a clinical level. However, the molecular pathways underlying long-COVID phenotypes are still poorly understood. ObjectivesThis study aims to cluster patients according to their blood transcriptomes and uncover the pathways underlying their disease. MethodsBlood was collected from 77 patients with long-COVID from the P4O2 COVID-19 study. Unsupervised hierarchical clustering was performed on the whole blood transcriptome. These clusters were analysed for differences in clinical features, pulmonary function tests and gene ontology (GO) term enrichment. ResultsClustering revealed two distinct clusters on a transcriptome level. Compared to cluster 2 (n=65), patients in cluster 1 (n=12) showed a higher rate of pre-existing cardiovascular disease (58% vs 22%), higher prevalence of gastrointestinal symptoms (58% vs 29%), shorter hospital duration during SARS-CoV-2 infection (median: 3 vs 8 days), lower Tiffeneau index (72% vs 81%) and lower diffusion capacity of the lung for carbon monoxide (DLCO) (68% vs 85% predicted). GO-term enrichment analysis revealed upregulation of genes involved in the antiviral innate immune response in cluster 1, while genes involved with the adaptive immune response were upregulated in cluster 2. ConclusionThis study provides a start in uncovering the pathophysiological mechanisms underlying long-COVID. Further research is required to unravel why the immune response is different in these clusters, and to identify potential therapeutic targets to create an optimized treatment or monitoring strategy for the individual long-COVID patient.