Background/Purpose: Vasculopathy, a hallmark clinical feature in systemic sclerosis (SSc), is a consequence of endothelial cell (EC) damage and dysfunction and might precede tissue fibrosis. Clinical disease progression in SSc is highly heterogeneous, with some patients developing severe fibrotic complications while others are affected by severe vascular complications including pulmonary arterial hypertension (PAH). The underlying mechanisms of this variability are poorly understood. We hypothesize that these distinct clinical phenotypes can be identified at the EC level. The objective is to investigate whether SSc patient plasma induces specific EC profiles and to examine correlations between these EC signatures and patients’ clinical features.

Methods: Human umbilical vein endothelial cells (HUVECs) were exposed to 25% citrate plasma from SSc patients from the Leiden CCISS cohort (n = 199; 61 at baseline, and 69 at baseline and 1-year follow-up) and healthy controls (HC; n = 35), according to previously published protocol (Postma RJ, 2024; doi:10.1016/j.gastha.2023.10.006). After staining, Max-projections of 9 z-steps with 0.7 µm step size were acquired using a high content confocal microscope. Single-cell morphological profiles were computed from the images and averaged to create single-patient profiles (Fig 1). The Leiden algorithm was used to detect densely connected clusters in the constructed dataset. Performance of Support Vector Machine models predicting different clinical characteristics was evaluated by repeated cross validation. Sankey plots were generated to visualize transitions between clusters after 1 year of follow-up. Validation experiments are pending.

Results: Graph-based cluster analysis of the morphological profiles identified 6 separate clusters, with high similarity between all 6-cluster solutions (adj. RAND index: 0.92). Cluster 6 represented outliers of the experiment and was therefore excluded from further analysis. HCs were predominantly found in cluster 0 (27.3%) and cluster 3 (55.9%). Figure 2 shows different SVM models that classify clinical features based on morphological data, specifically discriminating well between HC and patients (AUC 0.88 ± 0.09). Among patients, U1RNP+ patients and PAH patients showed distinct EC profiles (Fig 2). Strikingly, among patients EC profiles were clearly different between males and females, but this was not the case in HC (AUC 0.84 ± 0.08, and 0.68 ± 0.25, respectively). After 1 year follow up most patients (64.2%) migrated to a different cluster (Fig 3); patients who migrated from an SSc cluster to a HC enriched cluster (n = 8) showed stable, mild disease, whereas 50% of patients who migrated from a HC enriched to an SSc cluster (n=7/14) had disease progression.

Conclusion: Distinct morphological endothelial cell profiles were identified in SSc, especially for PAH. Interestingly, male and female EC profiles could be discriminated more clearly in SSc than in HC, indicating that sex-differences in SSc could be identified at the EC level. After 1 year follow-up, patients more often switched than remained in their original cluster, suggesting EC clusters could potentially reflect disease activity.

Figure 1. Morphological profiling on endothelial cell level.

Figure 2. ROC curves of cross validation of SVM models with clinical characteristics.

Figure 3. Transition of patients between cluster at baseline to cluster at 1-year follow-up.

« Back to ACR Convergence 2025

ACR Meeting Abstracts - https://acrabstracts.org/abstract/a-novel-endothelial-cell-profiling-assay-to-identify-systemic-sclerosis-patient-characteristics-at-the-endothelial-cell-level/