Fauvel, C., Webb, A., Liu, Z., Lin, S., Correa-Jaque, P., & Benza, R. L. “Pulmonary Arterial Hypertension risk assessment genomic model using Bayesian Network”.
BACKGROUND AND RATIONALE. Accurate risk assessment is essential to making individualized treatment decision in pulmonary arterial hypertension (PAH) patients. Yet, existing probabilistic risk assessment models are insufficient since they did not include contemporary genomic and imaging biomarkers.We aimed to build a risk assessment genomic model using Bayesian Network for PAH patients.
METHODS AND RESULTS. After performing a whole genome sequencing on 325 samples, variants were filtered for quality, assigned to genes, and filtered for function and population frequency. Retaining PAH patients that survived past 7 years or died prior to 5 years left 221 samples for analysis (mean age 54 years, 50% idiopathic PAH, 81% of female). Ingenuity Pathways Analysis was used to generate a list of pathways containing more than 1 mutated gene from our dataset. Thirty-one pathways which were significantly (fisher exact test p<0.05) associated with long- (>7 years) versus short-term survival (<5 years) were retained. Finally, a Bayesian Network (BN) model, showing interdependency (arrow direction) and association weight (arrow thickness) between the 31 selected pathways and length of survival was built. The 10-fold cross-validation AUC of this model averaged 0.75. Using the already published peer-reviewed articles, we were able to link each of the 31 pathways to the natural history of PAH including endothelial dysfunction, vascular smooth muscular cell proliferation, inflammation and dysimmunity, genetic and environmental factors, metabolism dysfunction and right ventricular effect (figure).
CONCLUSION. Using BN, we were able to provide the first PAH risk assessment genomic model including 31 pathways that may be related to the natural PAH course.