Interleukin 21, plasminogen activator inhibitor 1 and fractalkine improve risk stratification in elderly patients with pulmonary embolism

Abstract Background The importance of clinical parameters for predicting short-term mortality in hemodynamically stable patients with pulmonary embolism (PE) is acknowledged within the Pulmonary Embolism Severity Index (PESI) Score. Little is known about the value of novel biomarkers in this context. Given the pathophysiology of PE, markers related to inflammation and thrombosis, such as interleukin 21 (IL-21), fractalkine and plasminogen activator inhibitor 1 (PAI1) are of particular interest. Purpose To identify new protein biomarkers that improve 90d mortality prediction compared to the PESI score in elderly PE patients. Methods 1003 patients with venous thromboembolism aged ≥65 years were recruited between 2009 and 2011 from 9 hospitals in Switzerland (SWITCO65+ cohort). Among these, 778 patients were excluded due to missing data, deep venous thrombosis only, blood sample taken >1 day after diagnosis, no initial anticoagulation, systolic blood pressure <90 mmHg, withdrawal within 1 day or denying further use of data; 225 patients were prospectively included. Outcome was independently adjudicated up to 1 year after the index event. To identify candidate proteins, untargeted proteomic analyses (OLINK) were conducted in cases with 90d all-cause mortality (n=20) matched 1:2 (age, sex) to controls without events (n=40). 26 candidate proteins were then validated in all 225 PE patients using ELISA and Mesoscale analyses. Using ROC analyses, the most promising proteins to predict death within 90d were identified within this cohort. Optimal biomarker cut-off levels were derived using the Youden Index and survival analyses were computed. To evaluate the predictive value of these biomarkers, hazard ratios (HR) were calculated after adjusting for age, sex, and BMI. AUC-ROC analyses were performed to assess the biomarkers’ value for risk prediction compared to the PESI score. Results Baseline characteristics (Figure 1) did not differ between patients who died within 90d and those who did not, except for median PESI score (108 [95 – 125] vs. 87 [79 – 101], p<0.001). Optimal biomarker cut-off values were: IL-21: 56.17pg/ml, PAI1: 36.82pg/ml, fractalkine: 66’515pg/ml. Increased levels of IL-21 and PAI1 were associated with increased mortality (HR IL-21 = 1.5 [95% CI: 0.99 - 2.79], HR PAI1 = 2.6 [95% CI: 1.7 – 4.8]), whereas patients with elevated fractalkine levels were at lower risk to die within 90d (HR fractalkine = 0.37 [95% CI: 0.16 - 0.60]). For prediction of 90d mortality, the three assessed biomarkers showed a higher AUC than the PESI score. Combining the PESI score with the biomarkers did not further improve the AUC (Figure 2). Conclusions The protein biomarkers IL-21, PAI1 and fractalkine improve prediction of 90d mortality in elderly, hemodynamically stable PE patients compared to the PESI score. These biomarkers highlight the contribution of inflammation and thrombosis in the pathophysiology of PE and warrant further investigation.Baseline CharacteristicsROC Curves for Biomarkers and PESI Score