CARDIOLOGY/BEST AVAILABLE EVIDENCE

D-dimer as the Sole Screening Test for Acute Aortic Dissection:A Review of the Literature

Alexander Sutherland, MDJude Escano, DOTroy P. Coon, MD

From the 31st Combat Support Hospital, Baghdad, Iraq.

This article expresses the opinions of the authors and does not reflect the opinions of theDepartment of Defense.

From the Editor—Emergency physicians must often makedecisions about patient management without clear-cutdata of sufficient quality to support clinical guidelines orevidence-based reviews. Topics in the Best AvailableEvidence section must be relevant to emergencyphysicians, are formally peer reviewed, and must have asufficient literature base to draw a reasonable conclusionbut not such a large literature base that a traditional“evidence-based” review, meta-analysis, or systematicreview can be performed.

[Ann Emerg Med. 2008;52:339-343.]

ABSTRACT

Recent studies have proposed the use of D-dimer as ascreening tool to “rule out” acute aortic dissection, claiming asensitivity approaching 100%. We reviewed the literature todetermine whether D-dimer can be used as the sole screeningtool for acute aortic dissection. An Ovid MEDLINE search,1966 to present, was performed with the key words “aorticdissection,” “fibrin degradation products,” and “D-dimer,”limited to “human” and “English language.” Ten originalresearch articles were identified that directly addressed the use ofD-dimer in acute aortic dissection. There appears to be a subsetof patients with acute aortic dissection who develop intramuralhematomas, frequently found to have low or negative D-dimerlevels. In addition, none of the studies reviewed defined aspecific patient population eligible for D-dimer screening, andmany of the studies had wide sensitivity confidence intervals(CIs) because of low patient numbers. Despite a high sensitivity,D-dimer cannot be recommended as the sole screening tool foracute aortic dissection.

Ten per 100,000 Americans will have an acute aorticdissection, and missing the diagnosis may be catastrophic.1 Adefinitive diagnosis of aortic dissection largely relies onadvanced radiographic imaging, either invasive ornoninvasive. However, 38% of dissections are missed oninitial evaluation, and there are no validated clinical decisionrules for the clinical diagnosis of acute aortic dissection.2,3

Clinical suspicion and chest radiograph findings were theonly tools for determining which patients require further

imaging until recent studies proposed the use of D-dimer as

Volume , . : October

a screening tool for acute aortic dissection, claiming asensitivity approaching 100%.2,4-8 Our goal was to evaluatethe current literature for the use of D-dimer as the solescreening tool for acute aortic dissection.

SEARCH STRATEGIESAn Ovid MEDLINE (1966 to present) search was performed

with the key words “aortic dissection,” “fibrin degradationproducts,” and “D-dimer” (in combination), limited to“human” and “English language.” This search yielded 23research articles, and all study designs were eligible except casereports. The authors reviewed all citation abstracts, and onlyoriginal published research articles that addressed the use ofD-dimer as a diagnostic tool for acute aortic dissection wereincluded. Nine original research articles were identified thatdirectly addressed the use of D-dimer in acute aortic dissection.A 10th original research article was identified during manuscriptrevision by Journal Watch, which met inclusion criteria but wasnot yet listed in Ovid MEDLINE.8

When reviewing the available literature for the use of D-dimer in acute aortic dissection, we specifically looked forparticipant selection criteria for prospective studies; patientcharacteristics, including signs and symptoms for retrospectivestudies; timing of D-dimer use and the onset of symptoms;inclusion of ancillary test before D-dimer testing (ie, chestradiograph) and the use of a definitive diagnostic study fordiagnosis. Selection criteria evaluated for prospective studiesincluded factors or prediction rules used to define the suspicionof acute aortic dissection before D-dimer testing, reporting ofclinical factors used for enrollment, reporting of primaryadmission versus referrals, and clinical variables defining severityof illness on admission. The definitive diagnostic study wasconsidered valid if one of the following was used: transthoracicechocardiography, transesophageal echocardiography, autopsy,computed tomography (CT), magnetic resonance imaging(MRI), or angiography. The Table presents a summary ofreviewed studies.

ARTICLE SUMMARIESWeber et al4

This study combined results of both a retrospective and

prospective arm of patients with acute aortic dissection and took

Annals of Emergency Medicine 339

D-dimer Screening for Acute Aortic Dissection Sutherland, Escano & Coon

place in a tertiary referral center. Fourteen patients diagnosedwith aortic dissection who had a D-dimer level measured wereidentified retrospectively during a 5-year period (dates notdefined) and 10 patients were identified prospectively. Specificselection criteria were not provided. The investigators used alatex monoclonal antibody specific for D-dimer (Tina-quantassay; Roche Diagnostics, Mannheim, Germany), and a positivetest result was considered greater than 0.5 �g/mL. The authorsused transthoracic echocardiography, CT, MRI, transesophagealechocardiography, aortography, and autopsy to diagnose aorticdissection but did not specify the number of patients that werediagnosed by each method. All 24 patients with aortic dissectionhad increased D-dimer levels. Time from presentation to D-dimer sampling ranged from 1 hour to 120 hours. Overall, theauthors conclude that D-dimer testing could be useful inexcluding AD, reporting a sensitivity of 100% (95% CI 82.8%to 100%).

Perez et al5

This was a retrospective chart review of all emergencydepartment (ED) patients with a final diagnosis of acute aorticdissection. The study was performed at an academic urban adultLevel I trauma center from 1996 to 2000. ED charts wereidentified using International Classification of Diseases code441.0 (dissection of aorta). The presence of a D-dimer assay wasrequired for inclusion. Referred patients were excluded. Chartswere reviewed for demographics, medical history, signs andsymptoms at presentation, type of dissection, diagnostic tools,management, and outcome. The D-dimer assay used was asemiquantitative latex agglutination assay (Diagnostic Stago,Asnieres, France), and greater than 0.5 �g/mL was considered apositive result. CT, transesophageal echocardiography, or

Table. Summary of reviewed studies.

StudyNumber ofPatients

D-dimer Sensitivity,% (95% CI)

Cutoff,�g/mL*

Weber et al, 2003 24 100 (82.8–100) 0.5 TinaPerez et al, 2004 7 100 (56.1–100) 0.5 Sem

asEggbrecht et al, 2004 16 100 (75.9–100) 0.5 Qua

(DHazui et al, 2005 29 93 (75.8–98.8) 0.8 Late

diAkatsu et al, 2005 30 100 (85.9–100) 0.5 Rapi

(RHazui et al, 2006 113 92 (85.0–96.1) 0.4 Late

diOhlmann et al, 2006 94 99 (93.3–99.9) 0.4 ImmWiegand et al, 2007 25 88 (67.7–96.8) 0.5 ImmSbarouni et al, 2007 18 94 (70.6–99.7) 0.7 ElisaSodeck et al, 2007 65 100 (93.1–100) 0.1 STA-

98 (90.6–99.9) 0.586 (74.8–93.1) 0.9

*Raw data were not available to calculate sensitivity at a uniform cutoff level.

angiography was used for the diagnosis of aortic dissection in all

340 Annals of Emergency Medicine

patients. Seven patients met inclusion criteria, and all 7 had aD-dimer level greater than 0.5 �g/mL. No reasoning forordering a D-dimer could be determined from chart review.Sensitivity was 100% (95% CI 56.1% to 100%).

Eggebrecht et al6

This was a prospective study of 64 consecutive patients withchest pain, presenting within 48 hours of symptom onsetbetween 2002 and 2004. Sixteen of the 64 patients had acuteaortic dissection, 16 had acute myocardial infarction, 16 hadpulmonary embolism, and 16 had noncardiac chest pain. Onlypatients for whom onset of acute aortic dissection was clearlydetermined were included in the study. No other selectioncriteria were provided. Thirty-two patients made up a controlgroup of asymptomatic patients with chronic aortic dissection(defined as at least 14 days after onset on initial intense pain).The D-dimer assay used was a latex-enhanced turbidimetric test(D-dimer Plus; Dade Behring, Marburg, Germany); the upperlimit of normal was 0.25 �g/mL. All patients had aorticdissection confirmed by transesophageal echocardiography,aortography, CT, or MRI. The average time from symptomonset to presentation to the ED was 16�15.7 hours. Theauthors also reported a negative correlation between theabsolute D-dimer values and the time from onset of symptoms.All of the patients with acute aortic dissection had a D-dimervalue greater than 0.5 �g/mL, and the authors report asensitivity of 100% (95% CI 75.9% to 100%) with this cutoff.

Hazui et al9

This was a prospective study from Osaka MishimaEmergency and Critical Care Center, designed to define theoptimum cutoff value for D-dimer concentration for

D-dimer Assay

Numberof False

Negatives Study Design

t assay (Roche) 0 Prospective and retrospectivetitative latex agglutination 0 Retrospective

ive assay D-dimer plusehring)

0 Prospective and retrospective

lutination (Rochetic)

2 Prospective and retrospective

side cardiac assay)

0 Prospective

lutination (Rochetic)

9 Retrospective

ssay (Sta-Lia test) 1 Retrospectivessay (Lia test) 3 Retrospectiveas D-dimer) 1 Prospectiveagglutination (Roche) 0 Prospective

19

-quaniquansay

ntitatade Bx aggagnosd bedochex aggagnosunoaunoa(Vid

Latex

discriminating between acute myocardial infarction and acute

Volume , . : October

Sutherland, Escano & Coon D-dimer Screening for Acute Aortic Dissection

ascending aortic dissection. Twenty-nine consecutive patientswith acute aortic dissection and 49 patients with acutemyocardial infarction were enrolled between 2001 and 2003within 4 hours of the onset of symptoms. Ten patients wereexcluded as a result of receiving heparin before arrival. A latexagglutination D-dimer (Roche Diagnostic, Tokyo, Japan) andgreater than 0.8 �g/mL was considered positive. No specificenrollment criteria were reported. Aortic dissection wasdiagnosed with chest and abdominal enhanced CT. Twopatients with a thrombosed false lumen had a D-dimerconcentration of less than 0.8 �g/mL (raw data not provided).Overall sensitivity was 93.1% (95% CI 75.8% to 98.8%).

Akutsu et al7

This study was a prospective study of consecutive patientswith suspected acute aortic dissection who had a D-dimer leveldetermined by rapid bedside assay and were admitted during2002 and 2004. Suspicion of acute aortic dissection was definedas sudden onset of chest or back pain, with no definitive ECGfindings of myocardial ischemia. Acute aortic dissection wasdiagnosed with enhanced CT. Patients were divided into 2groups, those with and without acute aortic dissection. Patientcharacteristics evaluated included age, sex, time from onset ofsymptoms to D-dimer assay, blood pressure, and creatinineclearance. Two separate D-dimer assays were used, a bedsideRoche Cardiac D-dimer system and a latex agglutination assay(LIAS-AUTO-D-dimer; Kokusai-shiyaku, Hyogo, Japan).“Positive” was considered greater than 0.5 �g/mL for bothassays. There were a total of 78 patients with suspected acuteaortic dissection, and 30 were diagnosed with acute aorticdissection. The median time from onset to measurement of D-dimer was 4.5 hours, and the authors reported no correlationwith time and D-dimer level. All patients with acute aorticdissection had a positive D-dimer result, with a sensitivity of100% (95% CI 85.9% to 100%).

Hazui et al10

This was a single-center retrospective chart review of 113consecutive patients with acute aortic dissection who wereadmitted to Osaka Mishima Emergency and Critical CareCenter within 24 hours of symptom onset between 2001 and2005. Patients who had presented in cardiac arrest wereexcluded. Enhanced CT was the definitive study used fordiagnosis, and all patients had a latex agglutination D-dimerassay (Roche Diagnostic, Tokyo, Japan). A positive assay resultwas considered greater than 0.4 �g/mL. The investigatorsreviewed age, timing of D-dimer and onset of symptoms, sex,presence or absence of ascending aortic involvement,thrombosed false lumen present, length of dissection, presenceof shock, pericardial effusion, and renal function. One hundredfour of the 113 patients with acute aortic dissection had positiveD-dimer results. Eight of the 9 patients in this study with acuteaortic dissection and negative D-dimer results had thrombosed false

lumens (overall sensitivity 92%; 95% CI 85.0% to 96.1%).

Volume , . : October

Ohlmann et al2

This was a retrospective chart review aimed to determine thevalue of D-dimer for the diagnostic and prognostic evaluation ofpatients with acute aortic dissection. Investigators reviewed16,529 records between 1997 and 2003. Inclusion criteriaincluded a diagnosis of acute aortic dissection bytransesophageal echocardiography, CT, MRI, or autopsy lessthan 15 days since symptom onset and the presence of a D-dimer assay result. A control group of patients was also selectedduring the same period, including patients with “symptomsclinically suspicious of aortic dissection”; acute aortic dissectionhad been ruled out by either CT or transesophagealechocardiography, and a D-dimer test had been performed.Evaluated symptoms were not reported. The D-dimer used wasa Sta-Liatest D-DI immunoturbidimetric assay (DiagnosticaStago), and a positive result was considered greater than 0.4�g/mL. Ninety-four patients were identified and 93 had apositive D-dimer result. The one negative D-dimer result was0.3 �g/mL, and this patient had a parietal hematoma of theascending aorta, without an intimal flap. Overall, the authorsreport a sensitivity of 99% (95% CI 93.3% to 99.9%).

Wiegand et al11

This was a retrospective chart review of all patients with adiagnosis of aortic dissection who also had a D-dimer test resultavailable at the University Hospital in Basel between 2000 and2005. The LiaTest D-dimer immunoassay (Diagnostica Stago) wasused, and greater than 0.5 �g/mL was considered a positive result.All patients were diagnosed with transesophageal echocardiography,angiography, CT, or pathology. Twenty-five patients wereidentified, and 3 patients had a negative D-dimer test result. Noraw data were provided, but the authors found no correlationbetween onset of symptoms and timing of D-dimer test. Theauthors did not mention any signs or symptoms at presentation butfound no significance when comparing age, sex, extent ofdissection, or outcome. Authors did not state the type of dissectionin patients with acute aortic dissection and negative D-dimer result.Overall sensitivity was 88% (95% CI 67.7% to 96.8%).

Sbarouni et al12

This was a prospective study of 18 consecutive patientsdiagnosed with acute aortic dissection, admitted to OnassisCardiac Surgery Center, Athens, Greece. Inclusion criteria werethe presence of aortic dissection and a D-dimer assay result. AnElisa (Vidas D-dimer, Biomerieux, France) D-dimer assay wasused, and all patients were diagnosed with either CT ortransthoracic echocardiography/transesophagealechocardiography. The authors found no correlation betweensymptom onset and D-dimer. The type of dissection cannot bedetermined from the study, but the authors reported nodifference between Stanford A or B dissections and D-dimervalues. Overall, the authors report a sensitivity of 94% (95% CI

70.6% to 99.7%), using a cutoff of 0.7 �g/mL.

Annals of Emergency Medicine 341

D-dimer Screening for Acute Aortic Dissection Sutherland, Escano & Coon

Sodeck et al8

This was a prospective observational study of 65 patientswith Stanford A acute aortic dissection, presenting to a tertiarycare nontrauma ED (Vienna, Austria) between 2003 and 2007.D-dimer assays were measured immediately after diagnosis,using STA latex-agglutination test (Roche Diagnostics, Vienna,Austria). Three separate cutoff values were used for analysis,greater than 0.1 �g/mL, greater than 0.5 �g/mL, and greaterthan 0.9 �g/mL. The definitive diagnostic study used fordiagnosis is not provided in the article. The authors report themedian time from symptom onset to hospital admission as 4.8hours (interquartile range 2.4 to 16). No raw data for D-dimerlevels was reported, but D-dimer levels ranged from 0.24 to137.88 �g/mL (median 3.47 �g/mL). Sensitivity was reportedas 100% (95% CI 93.1% to 100%) using 0.1 �g/mL, 98%(95% CI 90.6% to 99.9%) using 0.5 �g/mL, and 86% (95%CI 74.8% to 93.1%) using 0.9 �g/mL as the cutoff.

THE BOTTOM LINEAccording to the evidence available, D-dimer cannot be

recommended as the sole screening tool for acute aorticdissection. There is a subset of patients with acute aorticdissection who develop a thrombosed false lumen, which maybe less likely to stimulate the clotting cascade than those withluminal extension, creating false-negative D-dimer results.2,9,10

Additionally, 4 of the 10 studies report a sensitivity of 94% orless, and 3 of the 5 studies with 100% sensitivity have the lowerrange of their 95% CIs less than 85%, which is not acceptablefor ruling out a disease with such a high mortality.4-6,9-12

Although the overall sensitivity for D-dimer in acute aorticdissection was high in all of the reviewed studies, no definitivepopulation is identified as eligible for screening. Two of the 10studies use either chest or back pain for inclusion but fail toprovide any other clinical data.6,7 In addition, the prevalence ofaortic dissection (14/64 and 30/78) in their populations was toohigh to assume that only chest pain or back pain was usedexclusively for inclusion.6,7 All of the remaining retrospectivestudies have inclusion bias and use the diagnosis of aorticdissection as inclusion criteria without providing any valuableclinical data.2,4,5,8,10-12 The 3 remaining prospective studies failto clarify or include selection criteria to determine the pretestprobability of any of the studied populations.

The reviewed literature also fails to answer many othercrucial questions about the use of D-dimer for aortic dissection.Although not all studies reported the time from symptom onsetto assay, the range was from 1 hour to 120 hours. With such alarge range and the nonspecificity of D-dimer, it is possible thatother disease processes falsely increased the assay result.Additionally, none of the studies discuss the ordering of anancillary test (ie, bilateral blood pressures, ECG, chestradiography) before drawing of D-dimer and how it influencedsuspicion or inclusion into the studies. It is also not clear fromany of the studies how clinical suspicion played a role intoinclusion. Finally, the severity of illness is not addressed in any

of the studies in relation to the ordering of D-dimer.

342 Annals of Emergency Medicine

Besides the studies by Hazui et al,9 no other studies use thesame D-dimer assay (both had the same first author). There isheterogeneity for the lower limit of detection among thedifferent D-dimer assays used, as well as the cutoff values usedamong the studies reviewed. This heterogeneity further limitsthe use of D-dimer for aortic dissection.

It is possible that clinical findings coupled with chest radiographfindings are the most important factor in diagnosis of acute aorticdissection and may perform better than D-dimer alone. A review in2002 evaluated the literature for clinical assessment of suspected acutethoracic aortic dissection. The authors found that several clinicalvariables—sudden onset of chest or back pain, tearing or ripping pain,migrating pain, pulse deficits/blood pressure deficits, neurologicdeficits, and enlarged or widened mediastinum on chest radiograph—may assist physicians in risk stratification.13 In any patient with high-risk clinical characteristics of acute aortic dissection or abnormalities onchest radiograph suggestive of dissection, it would be unreasonable torely on a negative D-dimer result when a definitive study or operationis needed. Furthermore, the nonspecificity of D-dimer, combined witha low-risk patient population, will likely lead to excessive advancedimaging. However, D-dimer may still have utility if it can be shown toincrease detection when used in combination with a validated clinicaldecision rule or other clinical characteristics of acute aortic dissection.

We do not believe it is safe to use D-dimer as the solescreening test for acute aortic dissection at any cutoff level.Future prospective research should focus on using clinicalvariables and ancillary studies in conjunction with D-dimer foracute aortic dissection.

doi:10.1016/j.annemergmed.2007.12.026

Supervising editor: Judd E. Hollander, MD

Funding and support: By Annals policy, all authors are requiredto disclose any and all commercial, financial, and otherrelationships in any way related to the subject of this article,that might create any potential conflict of interest. The authorshave stated that no such relationships exist. See theManuscript Submission Agreement in this issue for examplesof specific conflicts covered by this statement.

Publication date: Available online March 19, 2008.

Reprints not available from the authors.

Address for correspondence: Alex Sutherland, MD, 3426Bentwood Place, Highlands Ranch, CO 80126; 303-997-4411;E-mail aspst10@yahoo.com; alexsuth@hotmail.com.

REFERENCES1. Clouse D, Hallett J, Shaff H, et al. Acute aortic dissection:

population based incidence compared with degenerative aorticaneurysimal rupture. Mayo Clin Proc. 2004;79:176-180.

2. Ohlmann P, Faure A, Morel O, et al. Diagnostic and prognosticvalue of circulating D-Dimers in patients with acute aorticdissection. Crit Care Med. 2006;34:1358-1364.

3. Bushnell J, Brown J. Clinical assessment for acute thoracic aortic

dissection. Ann Emerg Med. 2005;46:90-92.

Volume , . : October

Sutherland, Escano & Coon D-dimer Screening for Acute Aortic Dissection

4. Weber T, Hogler S, Auer J, et al. D-dimer in acute aorticdissection. Chest. 2003;123:1375-1378.

5. Perez A, Abbet P, Drescher MJ. D-dimers in the emergencydepartment evaluation of aortic dissection. Acad Emerg Med.2004;11:397-400.

6. Eggebrecht H, Naber CK, Bruch C, et al. Value of plasma fibrinD-dimers for detection of acute aortic dissection. J Am CollCardiol. 2004;44:804-809.

7. Akutsu K, Sato N, Yamamoto T, et al. A rapid bedside D-dimerassay (cardiac D-dimer) for screening of clinically suspected acuteaortic dissection. Circ J. 2005;69:397-403.

8. Sodeck G, Domanovits H, Schillinger M. D-dimer in ruling outacute aortic dissection: a systematic review and prospectivecohort study. Eur Heart J. 2007 [Epub ahead of print].

doi:10.1093/eurheartj/ehm484.

Volume , . : October

9. Hazui H, Fukumoto H, Negoro N, et al. Simple and useful testsfor discriminating between acute aortic dissection of theascending aorta and acute myocardial infarction in the emergencysetting. Circ J. 2005;69:677-682.

10. Hazui H, Nishimoto M, Hoshiga M, et al. Young adult patientswith short dissection length and thrombosed false lumen withoutulcer-like projections are liable to have false-negative results ofD-dimer testing for acute aortic dissection based on a study of113 cases. Circ J. 2006;70:1598-1601.

11. Wiegand J, Koller M, Bingisser R. Does a negative D-dimer testrule out aortic dissection? Swiss Med Wkly. 2007;137:462.

12. Sbarouni E, Georgiadou P, Marathias A, et al. D-dimer and BNPlevels in acute aortic dissection. Int J Cardiol. 2007;122:170-172.

13. Klompas M. Does this patient have an acute thoracic aortic

dissection? JAMA. 2002;287:2262-2272.

IMAGES IN EMERGENCY MEDICINE(continued from p. 337)

DIAGNOSIS:Frontoethmoid mucocele. Computed tomography (CT) revealed a large frontoethmoid mucocele, with

significant proptosis of the left globe and inferior and lateral displacement of the eye (Figure 3). Thinning of thesuperior orbital roof, with erosion of the medial orbital wall, can be seen in Figure 4. The patient underwent aLynch frontal sinusotomy, removal of the mucocele, decompression of the orbit, and left endoscopicethmoidectomy. His visual changes resolved 2 days after the surgery, and the postsurgical CT 2 months laterrevealed decompression of the mass, with some displacement inferiorly of the globe. The patient had anuncomplicated postoperative course.

Mucoceles are slow-growing cystic lesions lined with respiratory epithelium containing mucous and serousfluid. Sinus mucoceles form when there is obstruction of the sinus ostium from congenital malformation, chronicsinus disease, infection, nasal surgery, allergies, facial trauma, or neoplasm. Pressure is then exerted by thenondraining mucocele, expanding the sinus and causing extension of the mass into important adjacent structuressuch as the orbit and cranial cavity. The most common location is frontoethmoid. Clinical variation can differ bylocation of the mass, such as frontal mucoceles which cause the eye to be pushed inferior, ethmoid mucoceles causeproptosis, and maxillary mucoceles can cause retropulsion. Infection of mucoceles is rare. Treatment is surgical,with endoscopic drainage being most favorable, although there are several surgical techniques.

REFERENCES1. Arrue P, Kany MT, Serrano E, et al. Mucoceles of the paranasal sinuses: uncommon location. J Laryngol Otol. 1998;112:

840-844.2. Wang T, Liao S, Jou J, et al. Clinical manifestations and management of orbital mucoceles: the role of ophthalmologists.

Jpn J Ophthalmol. 2005;49:239-245. .3. Yak SK, Aung T. Frontosinus sinus mucoceles causing proptosis: two case reports. Ann Acad Med Singapore.

1998;27:744.

Annals of Emergency Medicine 343