Improving risk stratification in acute pulmonary embolism: How do we target our therapies more effectively?

Acute pulmonary embolism (PE) remains a leading cause of cardiovascular morbidity and mortality across the globe. For the highest risk patients, mortality still nears 50% in contemporary cohorts with little improvement since the seminal ICOPER study published nearly two decades ago.^1,2 Progress has been modest despite a burgeoning number of new technologies and techniques utilized for the acute management of high-risk PE. While these therapies demonstrate potential for improved long-term outcomes and therefore have been welcomed by clinicians, there has been a lack of rigorous validation through randomized controlled trials. ¹

Appropriate risk stratification is a key component of the clinical pathway leading to targeted therapeutic options for a given patient; however, reliable and adequate risk stratification of acute PE has remained elusive. Several tools have been validated for this purpose, including the Pulmonary Embolism Severity Index (PESI), the simplified Pulmonary Embolism Severity Index (sPESI), and the BOVA score.^3–5 Another commonly utilized risk stratification schema is the 2019 European Society of Cardiology (ESC) and European Respiratory Society (ERS) guidelines on diagnosis and management of acute PE. ⁶ Under the ESC/ERS guidelines, patients are risk stratified into four strata based on a combination of clinical, radiographic, and laboratory assessments. These strata include low risk, intermediate-low risk, intermediate-high risk, and high risk for early mortality; however, a recent multicenter cohort study of over 400 patients demonstrated that this risk stratification paradigm and its four categories had modest discriminating ability for 7-day and 30-day mortality. ⁷ These findings suggest that our current risk stratification schemes are inadequate and desperately need improving.

In this issue of Vascular Medicine, Newman and colleagues report the utility of a novel biomarker, Angiopoietin-2 (Ang-2), in the risk stratification of patients with acute PE. ⁸ Ang-2 is a signaling molecule whose transcription is induced by hypoxic and inflammatory states and has been associated with worse prognosis in other acute cardiac conditions such as congestive heart failure. The authors designed a single-center prospective observational study, collecting samples at admission for patients with acute PE. Sixty-five patients were included in the cohort and tertiled into risk groups based on the ESC guidelines. The authors found an association between increasing levels of Ang-2 and PE severity, with the most pronounced difference being between intermediate and high-risk PE (3850 [IQR 3160–4773] vs 11,461 [IQR 7032–25,600] pg/mL, p = 0.03). Interestingly, there was no significant difference between Ang-2 tertiles and presenting heart rate or oxygen saturation. Mean arterial pressure decreased in a stepwise fashion with increasing Ang-2 tertile level. Invasive and noninvasive assessments of right ventricular (RV) dysfunction according to Ang-2 tertile did not differ significantly, save for pulmonary arterial systolic pressure (PASP), RV outflow tract velocity time interval (RVOT VTI), and RV acceleration time (RV AT). Of note, the right ventricular/left ventricular (RV/LV) ratio and tricuspid annular plane systolic excursion (TAPSE), the noninvasive measures of RV function most commonly associated with unfavorable prognosis in acute PE, did not differ significantly across Ang-2 tertiles in this cohort. ⁹ The authors created a receiver operating characteristic (ROC) curve for Ang-2 levels to evaluate the sensitivity and specificity of Ang-2 for predicting ICU admission (68% for both at a cutoff of 4101 pg/mL). In univariate analysis, Ang-2 correlated significantly with ICU admission, but correlation lost statistical significance under multivariate analysis. The authors then added Ang-2 to the sPESI score and compared ROC curves with and without the additional variable. They noted a relatively modest increase in the AUC from 0.638 to 0.682.

The authors should be commended on investigating a novel biomarker to improve risk stratification paradigms in acute PE. There has been little new development in this area over the last decade except for several papers that have updated existing data regarding cardiac troponins (with high sensitivity troponin-T assays) and B-type natriuretic peptides (with NT-proBNP assays).^10,11 We have established the importance of improving our risk prediction models, and Ang-2 is a welcome addition for further study; however, the findings must be placed in context.

Ultimately, Ang-2 had only a modest discriminating ability to estimate risk in acute PE in this report. It appeared to perform best in the highest risk patients but had difficulty in discriminating between low- and intermediate-risk patients. Further, there was no change in heart rate or oxygen saturation across tertiles of Ang-2. Similarly, invasive and echocardiographic parameters associated with higher risk PEs did not uniformly correlate with Ang-2. The modest discriminating capability, particularly at lower risk profiles, and inconsistent associations speaks to the heterogeneity of PE presentation. Although the medical system diagnoses acute PE at the time of admission, determining the actual length of time that thrombotic material has been accruing in the pulmonary arteries has proven difficult. It is possible that the percentage of patients with acute on chronic PE or subacute PE is higher than recognized, which leads to the varying degrees of hemodynamic compensation at presentation.

Given these inherent limitations combined with a limited sample size, the study by Newman and colleagues does not support routine use of Ang-2 in acute PE; however, several hypothesis-generating questions deserve further inquiry. How does Ang-2 perform in predicting hard clinical outcomes such as mortality? The authors mentioned a follow-up study, which will be eagerly anticipated. A relatively large proportion of patients received catheter-directed thrombolysis, which often necessitates an ICU admission. Would Ang-2 have been more strongly correlated with ICU admission if admission were strictly based on clinical criteria? A subgroup analysis to examine the Ang-2 levels in those admitted to the ICU without catheter-directed therapies may be helpful. Would Ang-2 have performed better in a larger or higher risk cohort? As a signaling molecule induced by tissue hypoxia, the mechanism of action suggests higher levels in patients with some component of end organ dysfunction. Should Ang-2 replace troponin or BNP as the risk stratification marker of choice in acute PE? This seems unlikely because the former two markers have numerous clinical applications outside of acute PE, but none of the biomarkers exhibited robust discrimination between risk tertiles in this cohort. How does Ang-2 perform in combination with troponin and/or NT-proBNP for risk stratification? Although not addressed in this report, it would be worth examining in subsequent studies to determine the additional discriminatory function of Ang-2 in combination with currently utilized biomarkers. Finally, is there a better combination of existing clinical variables and biomarkers that may improve risk stratification? If so, will improved risk stratification lead to better targeting of advanced therapies thereby leading to better clinical outcomes? Fundamentally, better clinical outcomes for patients with acute PE is the ultimate goal as mortality rates remain unacceptable for the highest risk patients. Continued research in both diagnostic and therapeutic strategies will hopefully provide effective mortality reduction in the future for these patients.