Prognostic significance of PaO2/PaCO2 ratio in normotensive patients with pulmonary embolism

Introduction: Risk stratification remains controversial in patients with normotensive pulmonary embolism (PE). The debate has recently focused right ventricular dysfunction detected by echocardiography or spiral computed tomography (CT) and cardiac biomarkers. Objectives: The utility of the PaO2/PaCO2 ratio to predict the short-term prognosis of PE is not currently known and that is the aim of the present study. Materials and Methods: This study retrospectively enrolled 99 (34 males, 65 females, 67 +/- 15 years) consecutive patients with acute PE, diagnosed by spiral chest tomography pulmonary angiography (CTPA). On admission, cardiac troponin T (cTn-T) was measured and on CTPA both right ventricle diameter and left ventricle diameter was calculated (RV/LV ratio). During the first 24 h after admission, all the patients had initial arterial blood gas collected under room air. Receiver-operating characteristic (ROC) analysis was performed to determine the optimal PaO2/PaCO2, RV/LV ratio and cTn-T cutoff level with regard to prognosis. Results: In-hospital mortality was 12.1% and all-cause 90-day mortality was 15.2%. Ten of 15 patients who died had a PaO2/PaCO2 <= 1.8 based on ROC analysis (P < 0.014). The cutoff level of PaO2/PaCO2 <= 1.8 had a high negative predictive value of 93% for mortality. Multivariable analysis revealed that PaO2/PaCO2 <= 1.8 Hazard Ratio (HR): 16.8 [95% CI: 2.6-108, P < 0.003] was the most significant independent predictor, whereas cTn-T, pO(2) < 60 mmHg and cardiac failure were nonsignificant factors. In addition, PaO2/PaCO2 <= 1.8 showed significant survival differences for overall mortality rates in Kaplan-Meier analysis (P < 0.012). Conclusion: The PaO2/PaCO2 measurement is a highly useful and practical measurement to predict prognosis in patients with acute PE. Moreover, it appears to be a more accurate predictor than RV/LV ratio and cTn-T levels in patients with normotensive PE.