A hemodynamically stable middle-aged male presents to your ED with
non-pleuritic chest pain and multiple cardiac risk factors. Your
diagnostic evaluation includes includes a Troponin-I which unfortunately returns at 0.5
ng/mL, although no ECG changes are noted. Unconvinced that this patient
has an acute coronary syndrome causing his chest pain, you perform a
pulmonary embolism protocol CT-angiogram and identify a PE. Although
you recall reading about a non-validated clinical decision rule (CDR)
to identify low-risk PE patients possibly safe for discharge home
(Aujesky D, et al. A Prediction Rule to Identify Low-Risk Patients with
Pulmonary Embolism, Annals Int Med 2006; 166: 169-175), you realize
this CDR is not yet ready for wide-spread use. More importantly, you
are uncertain how to interpret the unexpected Troponin elevation.
Diagnostic and prognostic studies relevant to Emergency Medicine decision making can be difficult to locate and even harder to read. As opposed to therapeutic articles which are increasingly summarized in well-designed systematic reviews like the Cochrane Collaboration, diagnostic articles are less frequently summarized in systematic reviews or meta-analyses. Furthermore, diagnostic study design methodology and quality assessment scores lag several years behind therapy-related research, further limiting one’s ability to confidently summarize findings across studies. Fortunately, you remember reading a prognostic article directly addressed the issue of abnormal Troponin levels in the setting of PE (Prognostic Value of Troponins in Acute Pulmonary Embolism: A Meta-Analysis, Circulation 2007; 116: 427-433).
This study was a meta-analysis of MEDLINE and EMBASE articles published between January 1998 and November 2006 with the primary objective to assess whether elevated troponins are associated with short-term mortality in acute pulmonary embolism patients. Individual articles were assessed for quality, heterogeneity, and publication bias. In addition, separate analyses were performed on retrospective and prospective studies as well as hemodynamically stable versus unstable subjects.
Among the twenty studies selected for inclusion, the mean age ranged from 53-69 years and most were predominantly female. Among those with an elevated troponin level (either troponin-I or troponin-T), 19.7% died versus 3.7% of those without a troponin elevation with an Odds Ratio 5.24 (95% Confidence Interval 3.28-8.38). Among those with a normal blood pressure, 17.9% with an elevated troponin died versus 2.3% mortality in those with normal troponin levels (OR 5.90; 95% CI 2.68-12.95). The specific troponin assay or cutoff point did not matter – all were associated with increased mortality when elevated. An elevated troponin level was also associated with an increased adverse outcome rate during the hospitalization.
This meta-analysis raises two questions: 1) How is an elevated troponin level pathophysiologically related to pulmonary embolism? 2) How is a clinician to use this information in their clinical decision making? The first question requires one to look at some of the meta-analysis details. Three of the studies included looked at echocardiography and troponin assays demonstrating an independent association of troponin elevation with right ventricular (RV) dysfunction. While a minority of these subjects may be suffering from an acute coronary syndrome in relation to their PE hemodynamic compromise, most are not likely suffering two concurrent problems. Rather, the sensitive troponin assay is reflecting RV strain. Since troponin is more readily available than echocardiography in the ED, the troponin level can be used as another marker for RV strain. This explanation answers the second question. Since RV dysfunction is associated with increased mortality with PE, the troponin elevation could be used as a risk-stratification tool to determine which hemodynamically stable PE patients require admission to a higher level of care for more intensive monitoring while therapeutic anticoagulation is obtained. While future health care systems may offer the possibility to discharge
home hemodynamically stable patients with acute PE, those with an elevated troponin will not be among those managed as outpatients due to their higher mortality and adverse event rates.
Bottom Line? Recognition of the high-risk, troponin-positive PE subset while in the ED can more effectively direct scarce resources at the population most likely to benefit from close monitoring, thereby reducing mortality and perhaps in-hospital adverse event rates.
Confidence Interval – a range of values within which the true value is likely to be found.
EMBASE – the European counterpart of MEDLINE with coverage ranging from 1974 to present and approximately 34% overlap of journals with MEDLINE. EMBASE is less biased towards English-language than MEDLINE, but is also proprietary and less accessible.
Heterogeneity – the differences or variability between studies with regards to their estimated treatment effects. For instance, in the troponin meta-analysis discussed above, different studies may generate different results due to differences in the study population, troponin assay used, and Gold standard for diagnosing PE.
MEDLINE – the world’s largest medical library with 11 million citations from over 4,000 biomedical journals and accessible for free (www.pubmed.gov).
Odds Ratio – the ratio of the probability of an event occurring to the probability of the event not occurring.
Publication Bias – the phenomenon of “positive” study results being published/presented/reported more commonly than “negative” findings. Good Systematic Reviews will assess for the possibility of unpublished negative reports through the use of an inverted funnel plot or through quantitative computational models.
Quality – in study assessment refers to minimized potential for bias. While multiple quality screening tools exist for different study types, the PE meta-analysis used a checklist proposed in a Systematic Review textbook (Egger M, Smith GD, Altman DG; Systematic Reviews in Health Care: Meta-analysis in Context, BMJ Books 2001, p 232) which assessed for patient-sampling, length of follow-up, objectivity of outcome assignment, prognostic variable precision and availability, analytical techniques, and standardized treatment descriptors.
Christopher Carpenter, MD, is an Assistant Professor at Washington University School of Medicine.