Cytokine Profiling of Systemic Sclerosis-related Pulmonary Hypertension.

Abstract

The study objective was to utilize multiplex immunoassays to create a cytokine profile specific for systemic sclerosis-related pulmonary hypertension (SSc-PH) by assessing circulating plasma of patients with SSc. This was an observational single-center study using plasma samples collected from 106 patients with SSc (24 with SSc-PH by right heart catheterization) in which we measured 48 circulating cytokines using multiplex immunoassays. We applied a random forest approach to create a full 48-cytokine model and a subsequent collapsed model based on variable importance by the mean decrease in the Gini index. The model was internally validated using a train-test split of the data from which we generated receiver operating curves and calculated sensitivity, specificity, and accuracy. The 48-cytokine model achieved an area under the curve of 0.82 (95% CI: 0.62, 1.00). Monokine induced by gamma interferon was the most important cytokine associated with SSc-PH by Gini index. Using the elbow method, a collapsed model was achieved using six cytokines (Monokine induced by gamma interferon, interleukin-8, macrophage inflammatory protein-1β, interleukin-15, interleukin-7, and macrophage-derived chemokine) that yielded an area under the curve of 0.83 (95% CI: 0.72, 1.00) with a sensitivity, specificity, and accuracy of 0.14 (95% CI: 0.00, 0.58), 0.96 (95% CI: 0.80, 1.00), and 0.78 (95% CI: 0.60, 0.91), respectively. Using a single-center SSc cohort that reflects real-world heterogeneity, we generated a highly specific cytokine profile for SSc-PH through a random forest approach. Future work will need to externally validate the cytokine profile through prospective cohorts.