June 17th, 2014
A Meta-Look at Thrombolysis vs. Conventional Anticoagulation for PE
Jay Giri, MD, MPH and Saurav Chatterjee, MD
CardioExchange’s John Ryan interviews Jay Giri and Saurav Chatterjee about their meta-analysis of trials comparing thrombolytic therapy and conventional anticoagulation in patients at intermediate risk for pulmonary embolism (PE). The study is published in JAMA. Further details are available in CardioExchange’s news coverage.
Ryan: Although you find a mortality benefit from thrombolytics, the number needed to treat (NNT) is 59. How do you discuss the potential benefits and harms with a patient when considering thrombolytic therapy?
Giri: As you suggest, clinical decision making about thrombolysis should be individualized on the basis of the risks and benefits for each patient rather than a one-size-fits-all approach to intermediate-risk PE. For example, both PEITHO, the largest randomized trial of this issue in intermediate-risk PE patients, and our analysis point to attenuated risks of bleeding in younger patients (75 or younger in PEITHO, 65 or younger in our meta-analysis). The most feared harm of thrombolytics, intracranial hemorrhage, appears to be significantly less likely in younger patients.
I favor a clear discussion with patients about whether to choose thrombolytic therapy for PE in selected cases. In most of those cases, patients are quite symptomatic and understand that a slightly increased risk for intracranial hemorrhage may accompany the possibility of more-rapid symptom resolution and other benefits, possibly even the mortality benefit suggested by our analysis.
Chatterjee: This is indeed a delicate issue. To aid clinicians in having such a discussion with patients, we outlined the different absolute-risk metrics in Table 2 of our article. We noted that for all patients presenting with PE, the NNT is 59 for preventing one death with thrombolytics, 78 for preventing one intracerebral hemorrhage (ICH), and 18 for preventing one episode of major bleeding (the definitions of which, unfortunately, varied across the studies, ranging from relatively more-common “drops in hemoglobin” to life-threatening ICH events). However, with effective thrombolysis, there is always a high risk for bleeding. To help make a balanced decision, we performed a formal, weighted risk–benefit analysis that showed, for all comers with PE who were potential candidates for thrombolytic therapy, a 0.81% weighted mortality benefit compared with the risk for ICH (an outcome comparable with the risk for death); this benefit persisted in the intermediate-risk population as well.
Ryan: RV dysfunction is defined in your paper as echocardiogram abnormalities and/or abnormal biomarkers (BNP, troponin). How useful do you find these parameters in determining treatment options for intermediate PE? (I find it difficult to justify thrombolytics in a hemodynamically stable patient.)
Giri: Prior research has shown that patients with echocardiographic markers of RV dysfunction or biomarkers indicating RV strain suffer worse outcomes than patients who don’t meet those criteria. Some people argue for treating those patients more aggressively with thrombolysis, but no trial to date had clearly justified that approach. Our analysis suggests that there is a group of hemodynamically stable patients who benefit acutely from thrombolysis.
The clinical factors I consider in making an individualized decision for a patient include the above parameters as well as heart rate, hypoxia, age, and bleeding risk. In intermediate-risk PE patients with advanced age or any increased bleeding risks (recent surgery/trauma, thrombocytopenia, history of stroke), I tend to take a more conservative approach, with anticoagulation alone as an initial strategy. However, for intermediate-risk patients who are younger without key relative bleeding contraindications, I consider thrombolysis to be a therapy that leads to faster symptom resolution, more-rapid normalization of pulmonary pressures and RV dysfunction, reduced chances of acute hemodynamic decompensation, and a potential mortality benefit. Of course, significant additional research must clarify quantitative risk-stratification strategies in this intermediate-risk cohort for both mortality and bleeding risks.
Chatterjee: This is an interesting question that merits further research. As you correctly note, RV dysfunction has been defined with various echocardiographic parameters (RV hypokinesis, RV dilatation, and pulmonary hypertension), as well as elevated biomarkers such as BNP and troponins. To our knowledge, the relative merits of the individual markers in identifying an optimal candidate for thrombolysis have not been determined. However, given our analysis of mortality benefits with thrombolytic therapy in intermediate-risk PE, any patients with PE who have features of RV dysfunction should at least be considered for thrombolytic therapy and should require a good reason for being precluded from a potentially life-saving therapy.
Ryan: Do you search for trials that are registered but unpublished, as on clinicaltrials.gov?
Giri: Yes, we used clincaltrials.gov and several other sources to search for unpublished data. In fact, in our original draft meta-analysis, the two most recent trials, PEITHO and TOPCOAT, had been presented but remained unpublished. However, during the peer-review process, those trials were published, leading our final accepted manuscript to include only published trials.
Chatterjee: We conducted a comprehensive search to identify all studies, including various databases and conference abstracts. In fact, we were able to identify two studies that had been reported at national conferences but had not made it to the publication stage by the time we first submitted our paper. During the JAMA peer-review process, the quality of the trials that had been presented but not published in peer-reviewed journals paved way for a debate on whether to include these studies in the final manuscript. Fortunately, the two trials were published in the interim, and a sensitivity analysis excluding them showed similar findings to those in the overall analysis, indicating robustness of the data (the sensitivity analysis was not included in the accepted final version). We did search clinicaltrials.gov to identify unpublished and ongoing studies, but we did not identify any controlled trials that would be reported in the near future and could further influence the results of our analysis.
Ryan: Is it proper to include so many older studies?
Giri: To avoid selection bias, we had our primary analysis incorporate all trials that met our inclusion criteria (randomization between anticoagulation and thrombolytics for acute PE). However, we agree that trials that are several decades old may not properly reflect the risks and benefits in contemporary practice. As such, we prespecified a secondary analysis of only the 6 most recent trials of thrombolysis (all published in the past 5 years and all including intermediate-risk PE patients). We discovered that the associated mortality benefit was strongly evident in this group of modern trials.
Chatterjee: This is a very valid issue, as many of the older thrombolytic trials were conducted in an era when standards of care were different. To prevent this issue from leading us to erroneous results, we performed a sensitivity analysis of the 6 trials published from 2009 to 2014 (since the last Cochrane review on this topic). We found that the mortality benefits actually were largely derived from these contemporary trials — a further reassuring message that should prompt clinicians to consider thrombolytics in PE. However, we did not want selection bias to affect our overall results, so we included all trials from all time periods for the overall analysis.
JOIN THE DISCUSSION
Do Giri and Chatterjee’s insights about their meta-analysis affect your view of the role of thrombolytic therapy for PE?
Absolutely. The meta-analysis reaffirms what we were actually doing daily practice.good to see unequivocal mortality benefits with thrombolysis.