Patrick Ohlmann, MD; Antoine Faure, MD; Olivier Morel, MD; H??e Petit, MD; Hasna Kabbaj, MD; Nicolas Meyer, MD; Edouard Cheneau, MD; Laurence Jesel, MD; Eric Epailly, MD; Dominique Desprez, MD; Lelia Grunebaum, MD; Francis Schneider, MD; Gerald Roul, MD, PhD; Jean-Philippe Mazzucotteli, MD; Bernard Eisenmann, MD; Pierre Bareiss, MD 

Crit Care Med.  2006;34(5):1358-1364.  ?2006 Lippincott Williams & Wilkins

Abstract and Introduction

Abstract

Objective: We sought to determine whether assessing D-Dimer might be helpful for the management of acute aortic dissection (AAD).
Design: Single-center retrospective case-control study.
Setting: University Hospital of Strasbourg France.
Patients: Patients were 94 consecutive patients admitted to our institution with confirmed AAD and in whom D-Dimer test had been performed at presentation. These patients were matched with 94 controls presenting with clinical suspicion of dissection, which was later ruled out.
Interventions: Patient characteristics and clinical course were analyzed.
Measurements and Main Results: Ninety-three (99%) patients with AAD had elevated D-Dimer (>400 ng/mL) with a median D-Dimer value of 8610 ng/mL (interquartile range, 2982-20,000 ng/mL). Receiver operating characteristic curves analysis showed that D-Dimer, but not C-reactive protein, troponin, lactate dehydrogenase, or leukocyte count, was predictive of a diagnosis of AAD, with a sensitivity and specificity of 99% and 34%, respectively. D-Dimer concentration positively correlated with the anatomical extension of the dissection to the different segments of the aorta (R = .47, p < .0001). A positive relationship was observed between D-Dimer and in-hospital mortality rate among patients with AAD (p = .037). On multivariate analysis, the independent predictors of in-hospital mortality were the presence of pericardial effusion (odds ratio, 6.80; confidence interval, 1.87-27.60), D-Dimer >5200 ng/mL (odds ratio, 5.38; confidence interval, 1.27-30.87), and female gender (odds ratio, 4.96; confidence interval, 1.39-19.95).
Conclusions: D-Dimers are elevated in patients with AAD and provide valuable diagnostic and prognostic information. In patients with acute chest pain and elevated D-Dimer, a diagnosis of AAD should also be considered. D-Dimer might be a useful complementary tool to the current diagnostic work-up of patients with suspected AAD.

Introduction

Acute aortic dissection (AAD) is a life-threatening cardiovascular emergency with a mortality rate of 1-2% per hour, early after symptom onset.^[1] Despite the introduction during the last decade of new imaging modalities including computed tomography (CT), transthoracic/transesophageal echocardiography (TTE/TEE), and magnetic resonance imaging (MRI), which allow accurate imaging of the disease, the mortality rate of AAD remains high.^[2]

Since almost 20% of patients may present without pain and with nonevocative symptoms like syncope, cerebrovascular accidents, or malperfusion syndrome of extremities or viscera or with congestive heart failure, evaluation of patients with suspicion of AAD is often difficult.^[1,2] Indeed, it was reported that diagnosis was missed in up to 38% on initial evaluation.^[3-5] Therefore, the main challenge in managing aortic dissection is to suspect and thus to diagnose the disease as early as possible.

Plasma D-Dimer, a degradation product of cross-linked fibrin, has been validated as a diagnostic tool to help in the exclusion of venous thrombosis and pulmonary embolism and is widely used in the emergency room setting.^[6-8] The gold standard methods of D-Dimer measurement are based on enzyme-linked immunoabsorbent assay technology,^[9,10] but processes have been developed to obtain faster results (in <15 mins) by fully automated immunoassay using immunoturbidimetric technology.^[11] Recently, elevations of circulating D-Dimers have been reported in patients with aortic dissection, suggesting that D-Dimer may be a biomarker of this disease.^[12-14] However, these studies have been conducted on relatively limited populations (<24 patients) and have not clarified whether D-Dimer is increased in the setting of aortic intramural hematoma^[15] and whether it may have a prognostic value in AAD. Therefore, the aim of this study was to determine, in a larger cohort, the value of D-Dimer measurements in the diagnostic and prognostic evaluation of patients with suspected acute aortic dissection.

Materials and Methods

Patients

Between January 1997 and December 2003, among 16,529 records of patients who were evaluated at our institution, we retrospectively identified 94 consecutive cases with a diagnosis of acute (i.e., <15 days after first symptoms) aortic dissection using the recommended criteria^[2] by imaging technique (TEE, CT, MRI) and/or autopsy, in whom a D-Dimer test had been performed as part of screening protocol at presentation. Because this was a retrospective study, neither informed consent nor ethics committee approval for the study was required under the French law. Dissections were classified on the basis of false lumen propagation to the ascending aorta (Stanford A and B and De Bakey 1, 2, and 3) as well as the parietal type of lesion (Swensson 1-5).^[16] Furthermore, the extension of the disease was quantified and reported in a scale from 1 to 6 based on the involvement of the following segments: ascending aorta, aortic arch, descending aorta, suprarenal abdominal, infrarenal abdominal, and iliac arteries.

Patients were matched with 94 controls admitted during the same period for evaluation because of symptoms clinically suspicious of acute aortic dissection, in whom AAD had been ruled out by CT or TEE and in whom a D-Dimer test had been performed. The final diagnosis in these patients was non-ST-segment elevation myocardial infarction in 12 cases (13%), ST-segment elevation myocardial infarction in 16 (17%), pulmonary embolism in 12 (13%), hemodynamic shock of undetermined origin in five (5%), heart failure in two (2%), aortic aneurysm without dissection in 15 (16%), pericarditis in seven (7%), esophageal/gastric disease in six (6%), neuroradicular pain in two (2%), and musculoskeletal pain in 17 (18%).

D-Dimer Test

D-Dimer level was assessed using the quantitative Sta-Liatest D-DI immunoturbidimetric assay (Diagnostica Stago, France) that measures the change in absorbance at 540 nm of a microlatex suspension coated covalently with two complementary monoclonal antibodies, specific for fibrin degradation products. The results collected are expressed in nanograms per milliliter. The range of detection of the assay is 220-20,000 ng/mL. The results are usually available within 15 mins. In our institution, the cutoff value is 400 ng/mL, with results above this threshold reported as positive. The agreement of this method with the standard enzyme-linked immunoabsorbent assay method is well established.^[17-19]

Statistical Analysis

Results are expressed as mean ? SD for continuous variables except for biological values that are expressed in median and interquartile range. Categorical variables are expressed as frequencies. Statistical analysis was carried out using Mann-Whitney test for continuous variables and Fisher's exact test for categorical variables. Correlations analysis was performed by Spearman rank test. Sensitivity, specificity, and positive and negative likelihood ratios with 95% confidence interval were calculated in relation to the final diagnosis, that is, having confirmed acute aortic dissection. Receiver operating characteristics (ROC) curves were constructed by plotting sensitivity (true-positive fraction) vs. 1-specificity (false-positive fraction) for predicting the diagnosis of AAD and for predicting in-hospital mortality. The area under the curve was calculated, and ROC curves were compared using the Hanley and McNeil method.^[20] To identify the independent predictors associated with in-hospital mortality among patients with AAD and due to the small number of events, exact univariate and multivariate ascending stepwise logistic regressions were used (LogXacT 4.1, Cytel Software Corporation). Odds ratios and their confidence intervals were determined. In all cases, p < .05 was considered to be significant.

Results

Patients' Characteristics

Baseline characteristics of the study patients are presented in Table 1 and Table 2 .

Diagnosis of AAD was confirmed by CT in 25 (27%), TEE in 24 (26%), TEE and CT in 38 (40%) MRI in two (2%), angiography in two (2%), and autopsy in three (3%) of cases. Dissections were classified as Stanford A in 67 (71%) patients and Stanford B in 27 (29%). Intramural aortic hematoma was observed in ten (11%) patients ( Table 2 ).

D-Dimer was measured after a mean duration of symptoms of 1.2 ? 2.5 days and before any surgical procedure. D-Dimer was elevated in 93 (99%) patients with aortic dissection. Median D-Dimer concentration was 8610 ng/mL (interquartile range 2982-20,000 ng/mL). In 26 (28%) patients, D-Dimer was >20,000 ng/mL. One patient with AAD and normal D-Dimer (300 ng/mL) had localized parietal hematoma of the ascendant aorta without intimal flap. One more patient with intramural hematoma had a D-Dimer level just above the cutoff value (430 ng/mL). D-Dimer was significantly lower (p < .0001) in patients with intramural hematoma (median 1230 ng/mL, interquartile range 685-2645 ng/mL) than in patients with patent false lumen (median value 9290 ng/mL, interquartile range 3890-20,000 ng/mL).

Correlation Between D-Dimer and Anatomical Extension of AAD

D-Dimer levels were correlated with the number of segment of dissected aorta (R = .47, p < .0001, Fig. 1) and tended to be higher in Sanford A than in Stanford B (medians 9260 ng/mL vs. 3975 ng/mL, p = .052). With respect to De Bakey classification, D-Dimer was higher in De Bakey I (median 15,705 ng/mL) than in De Bakey II and De Bakey III (medians 3050 and 3975 ng/mL, respectively, p < .05 for both vs. De Bakey I).

Figure 1. 

Box plot graph showing median, interquartile range, and 10th and 90th percentiles of D-Dimer concentration according to the number of aorta segments (1-6) involved in aortic dissection. The segments considered were ascending aorta, aortic arch, thoracic descending aorta, suprarenal abdominal aorta, infrarenal abdominal aorta, and iliac arteries. Nonparametric Spearman analysis showed very significant correlation between D-Dimer level and number of aorta segments involved in aortic (p < .0001) with r coefficient of .47.

     

Diagnostic Value of D-Dimer in AAD

In the control group, 62 patients (66%) had elevated D-Dimer. Mean D-Dimer level was significantly lower in control than in AAD patients ( Table 2 ). Individual values of D-Dimer in AAD and in control patients, divided by diagnosis, are presented in Figure 2A. ROC curve analysis showed that D-Dimer was highly predictive of a diagnosis of acute aortic dissection (area under ROC curve 0.88 ? 0.024, p < .0001, Fig. 2B). Sensitivity and specificity were 99% and 34%, respectively, at the usual cutoff value of 400 ng/mL ( Table 3 ). Furthermore, the positive and negative likelihood ratios were 1.5 and 0.03 ( Table 3 ). D-Dimers were not correlated with leukocyte count, C-reactive protein, or troponin I (data not shown).

Figure 2. 

A, scattergram showing the individual values of D-Dimer in acute aortic dissection (AAD) and in control patients divided by diagnosis. NSTEMI, non-ST-segment elevation myocardial infarction; STEMI, ST-segment elevation myocardial infarction; PE, pulmonary embolism; shock, hemodynamic shock; other, other diagnosis among controls (see Methods). B, receiver operating characteristic curves (ROC) for the prediction by D-Dimer level of acute aortic dissection. The area under ROC curve (AUC) ? sem and the 95% confidence interval are indicated. Some cutoff of D-Dimers (in ng/mL) values are shown on the curve.

     

Lack of Diagnostic Value of Other Biological Parameters for AAD

Data were available for analysis in 179 patients for leukocytes (87 control, 92 AAD), 151 for troponin I (76 control, 75 AAD), 108 for C-reactive protein (66 control, 42 AAD), and 102 for lactate dehydrogenase (52 control, 50 AAD). Leukocytes but not troponin I, C-reactive protein, and lactate dehydrogenase were slightly but significantly higher in patients with AAD than in controls ( Table 2 ). In Figure 3 are represented the separate values of leukocytes, troponin I, C-reactive protein, and lactate dehydrogenase in AAD and in control group patients divided by diagnosis. ROC curve analysis showed that only leukocyte count showed a positive diagnostic value for AAD (area under curve 0.61 ? 0.043, p = .009) with a sensitivity of 66% and a specificity of 46% at the usual cutoff value of 10⁹/L. When area under ROC curve of leukocyte count was compared with that of D-Dimer, it showed a significantly lower diagnostic values (difference between area 0.27 ? 0.046, p < .0001).

Figure 3. 

A, C, E, G, scattergrams showing the individual values of leukocytes (A), troponin I (cTnI, C), C-reactive protein (CRP, E), and lactate dehydrogenase (LDH, G) in acute aortic dissection (AAD) and in control patients divided by diagnosis. NSTEMI, non-ST-segment elevation myocardial infarction; STEMI, ST-segment elevation myocardial infarction; PE, pulmonary embolism; shock, hemodynamic shock; other, other diagnosis among controls (see Methods). B, D, F, H, receiver operating characteristic curves (ROC) for the prediction of acute aortic dissection by leukocytes count (B), cTnI (D), CRP (F), and LDH (H). The area under ROC curve (AUC) ? sem and the 95% confidence interval are shown for each graph.

     

In-Hospital Outcome in Patients With AAD

Twenty-two patients (23%) died during the in-hospital course (29% in type A and 11% in type B). The cause of death was hemopericardium with tamponade in seven patients (32%), aortic rupture in four patients (18%), and irreversible shock in two patients (9%). Death occurred in the perioperative phase in eight patients (36%), and the cause of death was unknown in one patient (4%).

Median D-Dimer was 15,920 ng/mL (interquartile range 5990-20,000 ng/m) in patients who died and 6600 ng/mL (interquartile range 1960-19,060 ng/mL, p = .033) in survivors.

ROC curve analysis for the prediction of death by D-Dimer (Fig. 4) revealed a significant link between D-Dimer level and in-hospital mortality among patients with aortic dissection (area under the curve 0.65 ? 0.066, p = .037).

Figure 4. 

Receiver operating characteristic curves (ROC) for the prediction by D-Dimer level of in-hospital mortality among patients with acute aortic dissection. AUC, area under ROC curve ? sem and the 95% confidence interval.

     

Univariate and Multivariate Analyses to Predict Mortality Among Patients With AAD

The logistic regression analysis performed to determine the significant predictors of mortality included the variables identified in the International Registry of Acute Aortic Dissection.^[1] Univariate analysis results are depicted in Table 4 . Factors associated with increased in-hospital mortality in univariate analyses were female gender, presence of pericardial effusion, systolic arterial pressure <100 mm Hg, presence of shock, and D-Dimer.

These variables were included in multivariate analyses. Three variables were identified in the final model to be independent predictors of mortality: presence of a pericardial effusion (odds ratio, 6.80; confidence interval, 1.87-27.60; p = .002), D-Dimer level >5200 ng/mL (odds ratio, 5.38; confidence interval, 1.27-30.87; p = .017), and female gender with (odds ratio, 4.96; confidence interval, 1.39-19.95; p = .011).

Discussion

Our data show that D-Dimer measurement has a high sensitivity in diagnosing AAD. Moreover, D-Dimer levels significantly correlate with the extension of the disease and are higher in patients with patent false lumen than in patients with intramural hematoma. Overall, a positive link between D-Dimer and in hospital-mortality was observed.

Until recently, laboratory tests played a minor role when assessing AAD but were considered to be useful to exclude other diseases.^[2] Elevations of C-reactive protein and leukocytosis have been described in AAD and may result from systemic inflammatory response.^[21] Additionally, lactate dehydrogenase may be raised when the celiac artery is involved in the dissection and may reflect visceral ischemia.^[2] In our study, none of these markers at presentation were predictive of the diagnosis of AAD. Suzuki et al.^[21] investigated the reliability of smooth muscle myosin heavy chain plasma levels in a series of 37 patients with AAD and reported a sensitivity and a specificity of 90% and 97%, respectively. However, smooth muscle myosin heavy chain is unavailable for routine practice. Recently, positive D-Dimer levels have been described in AAD^[12,14] in two small series (24 and 16 patients). Both found that D-Dimer had a 100% sensitivity to detect AAD and that D-Dimer level correlated with the time elapsed between onset of symptoms and blood testing.^[12,14] In a significantly larger series (94 patients), we have observed a sensitivity of 99%. Only one patient with a localized intramural hematoma had a normal D-Dimer (i.e., <400 ng/mL). Since data regarding the precise hour of symptoms onset were not available in the majority of the charts, we were unable to evaluate the relationship between D-Dimer levels and the duration of symptoms. However, in several patients we observed levels above the upper limit of the assay (20,000 ng/mL) as soon as 3 hrs after the beginning of symptoms, suggesting a very early increase of this biomarker in AAD.

In patients with AAD, D-Dimer was influenced by the anatomical extension and by the type of the dissection. The number of aortic segments involved in the dissection correlated with D-Dimer levels. This suggests that D-Dimer increases proportionally to the surface of contact between the bloodstream and the thrombogenic components of the false lumen. Furthermore, the lower level of D-Dimer observed in intramural hematoma compared with patent false lumen also supports this hypothesis.

D-Dimer measurement is currently used in the diagnosis of venous thromboembolic disease, where it shows low specificity and very high sensitivity when a cutoff of 400 ng/mL is chosen.^[6] In our study, 66% of patients in the control group showed elevated D-Dimer. This figure is in accordance with large trials evaluating D-Dimer in venous thromboembolic disease,^[22,23] in which the proportion of positive Sta-Liatest D-DI test among controls ranged from 56% to 68%, leading to a low specificity. Similarly, in our study, D-Dimer measurement had a low specificity (34%).

Our data indicate that the D-Dimer test has a high sensitivity and a low specificity at the usual threshold of 400 ng/mL for the diagnosis of AAD. We chose a low diagnostic threshold for the test, as in thromboembolic disease, since the cost of missing a diagnosis is high in AAD. Therefore, by using this threshold of 400 ng/mL, we showed that the D-Dimer test might consistently contribute, when negative, in the ruling out of AAD, since the negative likelihood ratio of the test is 0.03.

Of note, we observed that 38% patients with AAD ( Table 3 ) showed very high D-Dimer concentrations (>12,000 ng/mL) and in 28% D-Dimer was >20,000 ng/mL. Therefore, the positive likelihood of aortic dissection increased when D-Dimer was in the highest values (>12,000 ng/mL; positive likelihood ratio = 12). Consequently, in case of a high D-Dimer elevation, physicians should be aware of the possibility of AAD, even if the clinical presentation is nonevocative, and they should not only consider venous thromboembolic disease.

The mortality rate (23.4%) observed in patients with AAD in this study is similar to previous reports.^[24,25] We found a significant relationship between D-Dimer and in-hospital mortality. These results confirm a not significant trend observed in a previous study that included a lower number of patients.^[14] Since mortality rate is higher in Stanford A than in Stanford B, this relation might be the consequence of higher D-Dimer level in patients with Stanford A and might reflect the higher frequency of complications when the ascending aorta is involved. Furthermore, false lumen patency, which triggers the coagulation cascade and therefore possibly enhances D-Dimer level, is also known to be associated with a poorer prognosis,^[26,27] and might be involved in the relationship between D-Dimer and mortality. Overall, D-Dimer has been reported as a marker of risk for both multiple organ failure and death in critically ill patients,^[28] which may account for the higher mortality rate observed in patients presenting with the highest levels. Therefore, it is possible that patients with a D-Dimer level in the higher range are presenting with extensive forms of aortic dissection and thus have an increased risk of unfavorable outcome.

Study Limitations

The retrospective design of this study is the principal limitation of these findings. The control group was limited to matched patients and did not include all patients admitted for acute chest pain in our institution. However, we report the largest series of patients yet published with aortic dissection and D-Dimer testing. The immunoturbidimetric test was used for D-Dimer testing. Our results should not be extended to other D-Dimer assays.

Conclusions

Our data indicate that circulating D-Dimers, as assayed using an immunoturbidimetric test, are elevated in patients with AAD. D-Dimer levels are correlated with the anatomical extension of the disease and may also predict in-hospital mortality. Since the sensitivity of the test is high, the probability of AAD in patients with negative D-Dimer is low. In addition, the presence of a very high level of D-Dimer may be useful to suggest the diagnosis of AAD in patients admitted to the emergency room with nonevocative symptoms. D-Dimers may represent a complementary tool for the diagnostic and prognostic evaluation of AAD. Larger prospective studies should be conducted to confirm the role of circulating D-Dimers in AAD work-up and to define the accuracy of different D-Dimer assays in this setting.