D-dimer and Aortic Dissection
A 39 year-old, previously healthy, male presents to your emergency department with the chief complaint of chest pain. As the nursing staff gets the patient hooked up to the monitor, you learn from the patient that he experienced a sudden onset of a constant, ‘tearing type’ 6/10 pain in the substernal part of the chest that radiates to his shoulders and his back about 6 hours ago without appreciable aggravation or alleviation. The patient denies shortness of breath, nausea, vomiting, fevers, chills, cough, history of cardiac disease, history of DVT/PE, unilateral leg swelling or posterior calf pain. Patient has no appreciable past medical or surgical history. He takes no medication regularly. For social history, the patient is married, father of 3 children, works out regularly and denies history of smoking, illicit drugs or excessive alcohol use.
On exam you note a somewhat comfortable, normal heighted male with a blood pressure of 141/89, heart rate of 61, oxygen saturations of 100% and non-labored respirations at a rate of 16 breaths per minute. Exam is remarkable for clear lung sounds, regular rhythm with no murmurs, and equal 2+ radial pulses. There is no evidence of lower leg swelling or posterior calf pain. An electrocardiogram is handed to, as per your emergency department protocol, which demonstrates normal sinus rhythm without evidence of ischemia or arrhythmia. You quickly run through a differential diagnosis and consider, among other things, aortic dissection.
The Low Risk Patient
Aortic dissection is a longitudinal tearing of the wall of the aorta, which produces a false lumen that often propagates. Aortic dissection is a rare, but deadly pathologic process. The incidence is 3 in 100,000 people per year. Of every 10,000 emergency department visits, aortic dissection is only seen about once. Mortality for untreated proximal aortic dissection is about 25% by 24 hours of initial presentation and 80% at two weeks. Furthermore, clinicians are not good at detecting aortic dissection. The diagnosis is missed in up to 38% of patients on their initial evaluation and unfortunately, 28% of the time, the first diagnosis of dissection is made on post mortem examination . Although CT, TEE, and MRI are all common ways to evaluate for dissection, it would appear that spiral CT with contrast and appropriate electrocardiogram gating is the most sensitive imaging with sensitivity and specificity approaching 100%.
So how does one approach the patient in which aortic dissection is considered but has low risk or low pre-test probability? Furthermore, what constitutes low risk or low pre-test probability? The Aortic Dissection Detection (ADD) Risk Score was developed to address the latter question. The ADD Risk Score was put forth as part of a joint guidelines between 10 professional societies (emergency medicine not included) which is a total of 12 risk markers divided among 3 categories: predisposing conditions, pain features, and physical findings. Risk score ranges from zero to greater than one depending on how many of the categories the patient has risk markers present in. A score of zero was found in 4.3% of aortic dissections in a retrospective analysis of analysis of the International Registry of Acute Aortic Dissection done by Rogers et al. Further validation was performed by Nazerian et al with a prospective study with 1328 patients who aortic dissection was suspected in. In this study, if the ADD was 0 (or low risk), the prevalence was about 6%. The overall prevalence of dissection was very high in this study, however, at about 22%.  Is 4-6% an acceptable miss rate in aortic dissection? Probably not given that the outcome of dissection left untreated is such a devastating disease process. For comparison, a low-probability Well’s Score is associated with 2-4% prevalence of pulmonary embolism .
|Aortic Dissection Detection Score|
|Predisposing Conditions||Pain Features||Physical Exam Findings|
Family history of aortic disease
Known aortic valve disease
Recent aortic manipulation
Known thoracic aortic aneurysm
|Abrupt Onset of Pain
Severe Pain Intensity
Ripping or tearing pain
|Pulse deficit or SBP differential
Focal neurological deficit + pain
New Aortic Insufficiency murmur + pain
|If positive in any column, then add one point
Score=0; low risk (4.3%)
Score=1; intermediate risk (36.5%)
Score >1: high risk (59.2%)
|Adapted from Rogers et al. |
The Role of D-dimer in Dissection
Before diving into the discussion on D-dimer and aortic dissection, it should be noted that there are many studies and even many meta-analyses regarding this topic. This is probably an indicator that there is no decisive study to define the role of D-dimer in aortic dissection. That being said, there likely is a role for D-dimer, if somewhat limited, in the evaluation of aortic dissection.
Shimony et al conducted a meta-analysis of D-dimer in aortic dissection including 734 patients across seven studies which demonstrated plasma D-dimer less than 500 ng/mL was 97% sensitive with a negative LR of 0.06 . There are at least four other meta-analyses and systematic reviews that demonstrate similar results [9-12]. Pitfalls of these meta-analyses and systematic reviews and Shimony et al are that the studies that are included are primarily retrospective cohorts sprinkled with case series and case-control studies and the studies lack any uniform analysis of D-dimer. Recently a meta-analysis was published in Annals of Emergency Medicine by Asha and Miers regarding D-dimer and aortic dissection. This paper included four prospective studies with a total 1,557 patients and a cutoff of 500 ng/mL for a negative D-dimer result and concluded a sensitivity of 98% and a negative LR of 0.05. This paper also had pitfalls including an inherit selection bias, eliminating a third of the participants of the largest study because that particular study did not have D-dimers reported and all of the studies included in this meta-analysis had incidences of dissection between 23% to 48%.
Applying the D-dimer
How can this test be applied to the patient with concern for aortic dissection? If one uses the pre-test derived from the ADD scoring system and a negative LR of 0.05 derived from the discussed meta-analyses, it is easy to see that the D-dimer is probably only good enough for the low pre-test probability patients. So again, the question is who is low risk? Recalling the validation of the ADD Risk Score, low risk patients were defined as an ADD score of 0. The biggest issue with this is that likely any patient that one is considering dissection in, likely has one of the 12 risk factors in the ADD placing their risk score >0 automatically making them intermediate risk. For instance, the clinical vignette above describes a person if not for the characteristic of the pain; aortic dissection would not seriously be in the differential. There is a very small group of patients who there is concern for aortic dissection and who have an ADD scoring system of 0.
Gorla et al tackled this question with “Accuracy of a diagnostic strategy combining aortic dissection detection risk score and D-dimer levels in patients with suspected acute aortic syndrome”. In this paper, ‘low probability’ was defined as an ADD less than or equal to one. In this study, 376 patients were evaluated for dissection. Of these, 319 patients had an ADD score less than or equal to one and 57 patients had a ‘high probability’ with ADD score greater than one. Of those in the 319 patients in the ‘low probability’ group with D-dimers less than 500 ng/dL, none were found to have dissection but two were found to have aortic ulcers. This left the ‘low probability’ plus negative D-dimer with a sensitivity of 98.7% and NPV of 99.2%. Probably the largest pitfall of this study was that there was an incidence of 23% for aortic pathology which is much higher than most literature suggests.
Pitfalls of D-dimer in Aortic Dissection
These factors in acute aortic dissection are known to cause a false negative D-dimer test:
The bottom line is that D-dimer can likely be used in a select low pretest probability group for aortic dissection. The real question is who is low pretest probability? It is still up to the clinician and their gestalt as well as how comfortable they are deciding that a patient with an ADD score of 1 is low pretest probability. Furthermore, indiscriminate use of D-dimer for any chest pain will result in more unnecessary radiological testing.
Return to the Vignette
Decision is made that patient is not low probability for dissection given ADD score of 1 and CT scan was ordered which was negative for dissection. Patient dismissed after 3 hour troponin was negative, his pain had resolved, and lack of emergent pathology was discovered.
Currently, there is no good quality evidence to support the use of risk stratification models or to adopt the use of D-dimer in low-risk patients; as stated by the current ACEP clinical guideline in the topic. (Editor's Note)
Image by Steve Rotman (flickr) under CC BY-NC-ND-2.0