Peripheral Blood Transcripts Predict Preoperative Obstructive Total Anomalous Pulmonary Venous Connection

Abstract

The lack of accessible noninvasive tools to examine the molecular alterations limits our understanding of the causes of total anomalous pulmonary venous connection (TAPVC), as well as the identification of effective operational strategies. Here, we consecutively enrolled peripheral leukocyte transcripts of 26 preoperative obstructive and 22 non-obstructive patients with TAPVC. Two-hundred and fifty six differentially expressed mRNA and 27 differentially expressed long noncoding RNA transcripts were dysregulated. The up-regulated mRNA was enriched in the hydrogen peroxide catabolic process, response to mechanical stimulus, neutrophil degranulation, hemostasis, response to bacterium, and the NABA CORE MATRISOME pathway, all of which are associated with the development of fibrosis. Furthermore, we constructed predictive models using multiple machine-learning algorithms and tested the performance in the validation set. The mRNA NR3C2 and lncRNA MEG3 were screened based on multiple iterations. The random forest prediction model can predict preoperative obstruction patients in the validation set with high accuracy (area under curve = 1; sensitivity = 1). These data highlight the potential of peripheral leukocyte transcripts to evaluate obstructive-related pathophysiological alterations, leading to precision healthcare solutions that could improve patient survival after surgery. It also provides a novel direction for the study of preoperative obstructive TAPVC.