Biomarker-based in-hospital risk prediction for patients with intermediate risk pulmonary embolism

Background: Pulmonary embolism is a frequent and potentially lethal disease. Currently, risk prediction is primarily based on the ESC risk score, which categorizes patients in three risk categories (low, intermediate, and high). However, patients in the intermediate risk category pose a clinical challenge, as they frequently deteriorate despite up-front treatment with anticoagulants, which in part is due to great heterogeneity in this risk category. Ultimately, the ESC risk score lacks granularity to distinguish patients which can be solely treated with anticoagulants and those who would benefit from more invasive treatments early on.With this project, we aim to explore if established biomarkers can enhance in-hospital risk prediction for patients with intermediate risk pulmonary embolism.

Methods: Consecutive patients presenting with intermediate or high risk pulmonary embolism to a large tertiary care hospital in 2021 and 2022 were retrospectively enrolled. Two independent physicians adjudicated the ESC risk score, collected data on established (and routinely measured) biomarkers as well as the primary endpoint of hemodynamic worsening (defined as a need for inotropes, catecholamines, invasive ventilation or mechanical circulatory support) or in-hospital death. Sex- and age-adjusted logistic regression models evaluating the association between the measured biomarkers (NTproBNP, high-sensitive Troponin I, lactate, creatinine, ASAT, C-reactive protein, and INR) at baseline as well as their 24-hour delta and the primary endpoint were fitted.

Results: N=374 patients with acute intermediate risk pulmonary embolism treated in 2021/2022 were enrolled, median age 65 (interquartile range 57-75) years, 56% male. Mean baseline NTproBNP was 3794 pg/ml, with a mean change of -279 pg/ml within 24-hours. Baseline high-sensitive Troponin I was 296 pg/ml, with a mean change of -5 pg/ml within 24-hours. Baseline lactate was 1.6 mmol/l, with a mean change of -0.2 mmol/l within 24-hours. Baseline creatinine was 1.16 mg/dl, with a mean change of -0.12 mg/dl within 24-hours. Baseline ASAT was 88 U/l, with a mean change of -31 U/l within 24-hours. Baseline C-reactive protein was 80 mg/l, with a mean change of +6.3 mg/l within 24-hours. Baseline INR was 1.1, with a mean change of -0.03 within 24-hours. The primary endpoint of hemodynamic worsening or in-hospital death occurred in 32 patients (8.5%). After adjusting for age and sex, only baseline C-reactive protein (odds ratio 1.86, 95% confidence interval 1.24-2.78, p<0.01) and baseline INR (odds ratio 5.6, 95% confidence interval 1.19-26.39, p=0.02) were significantly associated with a higher likelihood of the primary endpoint.

Conclusion: In this single-center, retrospective analysis of consecutive patients presenting with acute intermediate risk pulmonary embolism, the established biomarkers C-reactive protein and INR enhanced in-hospital risk prediction. These results highlight that inflammation and coagulation, but more likely right heart function, play a major role in the pathogenesis of pulmonary embolism. Follow-up studies will show whether this association will primarily be used for diagnostic purposes, or if they can unravel new adjunctive treatments for patients with pulmonary embolism.

Figure 1: Cohort density of C-reactive protein and INR as well as their association with hemodynamic worsening and in-hospital death in N=374 patients with acute intermediate pulmonary embolism.