Letter by Hugli Regarding Article, “Diagnosis of Acute Aortic Dissection by D-Dimer: The International Registry of Acute Aortic Dissection Substudy on Biomarkers (IRAD-Bio) Experience”
To the Editor:
The authors of the study must be commended for their valuable efforts to help clinicians rule out aortic dissection (AD) with a biological marker as readily available as D-dimers.1 The exclusion of AD on the basis of a D-dimer level below a predetermined cutoff is an enticing replication of the diagnostic strategy developed to rule out pulmonary embolism, another cause of chest pain characterized by intraluminal thrombosis. The safety of the pulmonary embolism strategy relies on 3 sequential steps, each one validated by numerous studies: (1) use of clinical decision rules or clinical gestalt allows reliable assessment of the pretest probability of pulmonary embolism; (2) a precise estimate of the negative likelihood ratio of D-dimers leads to precise posttest probabilities of pulmonary embolism; and (3) clinical outcome studies have demonstrated the safety of this rule-out strategy.2
None of these steps have been validated for AD so far. First, pretest probability cannot be estimated by a clinical prediction rule; clinical gestalt may be insufficient because AD mimics many other diseases or is even asymptomatic in up to 10% of cases. The exclusion of AD on the basis of normal D-dimer levels is hazardous unless an accurate pretest probability is factored in. For example, a pretest probability of AD of 40% (the prevalence of AD in this study), coupled with the published negative likelihood ratio of 0.07, leaves a posttest probability of 4.5%; if the pretest probability increases only slightly to 60%, the posttest probability jumps to 9.5%, an unacceptably high residual risk in the face of a disease with such high mortality. Second, the negative likelihood ratio of D-dimers in the present study is most likely too small. The spectrum of AD consisted of a higher proportion of patients with a classic clinical presentation because they were recruited in tertiary care centers and were included only if their clinical presentation was suggestive enough of AD to undergo the diagnostic imaging test. This spectrum bias is reflected by the 40% prevalence of confirmed AD, much higher than the 0.3% of AD found in patients with chest pain presenting to an emergency department where the test is most likely to be used.3 In addition, it is also unclear from the International Registry of Acute Aortic Dissection Substudy on Biomarkers (IRAD-Bio) description that selection bias was avoided by recruiting consecutive patients, as was verification bias by blinding physicians who ordered the imaging test to the D-dimer result. These biases would also lead to an overestimation of test performance.4 Therefore, studies using a rigorous reporting methodology, as outlined in the Standards for Reporting of Diagnostic Accuracy (STARD) statement, are needed to confirm the results of this study.5 Finally, prospective clinical trials need to validate the safety of this rule-out strategy when it is generalized to clinical settings with a lower prevalence of AD and a wide spectrum of AD clinical presentation.
Until then, it would be unwise to rely solely on normal D-dimers to discharge patients suspected of having AD, however tempting the simplicity of this strategy may appear.
Suzuki T, Distante A, Zizza A, Trimarchi S, Villani M, Salerno Uriarte JA, De Luca Tupputi Schinosa L, Renzulli A, Sabino F, Nowak R, Birkhahn R, Hollander JE, Counselman F, Vijayendran R, Bossone E, Eagle K. Diagnosis of acute aortic dissection by D-dimer: the International Registry of Acute Aortic Dissection Substudy on Biomarkers (IRAD-Bio) experience. Circulation. 2009; 119: 2702–2707.