Biomarkers model for prediction of hospitalisation duration and pulmonary embolism outcome

Abstract

Introduction: Risk stratification is of the utmost importance for the management of patients with pulmonary embolism (PE) as it can improve the survival rate. Currently used, risk prediction models involve clinical, laboratory and echocardiographic parameters.

Aim and Objectives: To define a biomarker model as a prognostic tool for hospitalisation duration of PE and outcome, with similar or better accuracy compared to preexisting risk models.

Methods: Patients admitted for an acute PE were analyzed prospectively and routine clinical and laboratory values were measured. The outcome was determined by hospitalisation duration, thrombolysis, death and composite outcome (thrombolysis or death).

Results: We included 116 patients (38.8% male) with a median age of 53.5 (IQR 48-58). The composite outcome occurred in 20 (17.2%) patients. Multivariate and ROC curve analysis revealed that a model combining glucose and D-dimers showed the highest prognostic accuracy for mortality [AUC=0.867 (95% CI 0.745-0.989), p<0.001] and composite outcome [AUC=0.809 (95% CI 0.679-0.940, p<0.001)]. A significantly better prognostic performance compared to PESI score, simplified PESI score and European Society of Cardiology (ESC) risk classification for mortality and marginally better for the composite outcome was also shown. In multivariate regression analysis, strong predictors of hospitalisation duration were the biomarkers model [HR=4.148 (95% CI 1.672-6.624, P=0.001)] and male gender [HR=1.380 (95% CI 0.430-2.330, P=0.005)].

Conclusions: A biomarker model that included d-dimers and glucose had a better prognostic performance for PE outcome compared to ESC risk classification, PESI and simplified PESI score.