Related Experiment Video
Updated: Oct 21, 2026
In Vitro Microfluidic Disease Model to Study Whole Blood-Endothelial Interactions and Blood Clot Dynamics in Real-Time
Published on: May 24, 2020
Integration of hereditary thrombophilia data and polygenic risk scores into prognostic models improves differential
Yaroslav Kubkin1, Dmitry Nikulin2, Galina Okhrimenko1,3
1Centre for Biomedical Research and Technologies, Institute of Artificial Intelligence and Digital Sciences, HSE University, Moscow, Russia.
Background:
Differential diagnosis between pulmonary arterial hypertension (PAH) and chronic thromboembolic pulmonary hypertension (CTEPH) is crucial for treatment selection, particularly regarding surgical intervention. The prevalence of hereditary thrombophilias in Russian CTEPH patients and the utility of genetic data in differential diagnosis remain understudied.
Objectives:
To evaluate the prevalence of inherited thrombophilia among patients with CTEPH and PAH, to benchmark the results against published data from other cohorts, and to assess the potential utility of integrating genetic findings and polygenic risk score (PRS) parameters with clinical data for differentiating CTEPH from PAH.
Methods:
The study included 390 patients (252 PAH, 138 CTEPH). Whole-genome sequencing was performed to analyze classical thrombophilia genes (PROC, PROS1, SERPINC1, F2, F5, ADAMTS13, HRG, JAK2, PIGA, THBD, F9, FGA, FGB, FGG, CALR) and calculate VTE polygenic risk scores. A meta-analysis of 15 studies on hereditary thrombophilia prevalence in CTEPH was conducted. AutoML algorithm was used to construct diagnostic models.
Results:
Rare variants in thrombosis-associated genes were significantly more prevalent in patients with CTEPH than in those with PAH (12.3% vs. 1.2%, p < 0.001). Similarly, carriers of the common F5 and F2 thrombophilia variants were more frequent in the CTEPH group (15.9% vs. 6.3%, p = 0.002). Meta-analysis yielded pooled prevalence estimates in CTEPH of 3.0% for antithrombin deficiency, 3.6% for protein S deficiency, 4.7% for protein C deficiency, 7.9% for F5 Leiden, and 5.6% for F2 G20210A; these estimates were consistent with those reported in other populations. The clinical model yielded an ROC AUC of 0.883. The addition of PGS004854 and rare variants in thrombosis-associated genes increased the AUC to 0.951, representing an absolute increase of 0.068 (7.7%; p = 0.026). In the multivariable model, PGS004854 was the strongest positive genetic predictor of CTEPH (β = 1.869, p < 0.001).
Conclusions:
The genetic profile of CTEPH patients differs significantly from PAH, with a higher frequency of hereditary thrombotic risk factors. Integrating clinical data with polygenic risk scores and thrombophilia variants improves differential diagnosis accuracy and may optimize the clinical workup.