sinaiem.org

There’s a certain symmetry to starting the first journal club of the year studying d-dimer for dissection (last time around, we looked at the PERC rule and d-dimer for PE). This month, Bing examined a paper by Ohlmann and others, called “Diagnostic and prognostic value of circulating D-Dimers in patients with acute aortic dissection” (Crit Care Med. 2006 May;34(5):1358-64. PMID: 16557157). For background, we also read “D-dimer in ruling out acute aortic dissection: a systematic review and prospective cohort study” (Eur Heart J. 2007 Dec;28(24):3067-75. PMID: 17986466) and got some perspective on D-dimer the with Klompas addition to the JAMA rational clinical exam series: “Does This Patient Have an Acute Thoracic Aortic Dissection?” (JAMA 2002;287(17):2262-2272. PMID: 11980527).

The clinical background was useful in delineating the scope of the problem: acute aortic dissection (AAD) is rare — in some rigorous European surveys, incidence is 3-4 / 100,000 and in EDs, 2-3 of 1000 chest pain patients have AAD (so if you see about 1000 chest pain patients in the course of your residency…). We miss it a lot — Klompas said 39% of patients have a delay in diagnosis of more than 24 hours, 10% of autopsies for show missed dissections, and we only suspect AAD properly less than half the time (as low as 15% of the time).

There are a few key features of dissection that should set off alarm bells for AAD — pulse deficits and blood pressure cuff differences > 20 mmHg (positive likelihood ratio 5.7), focal neuro deficits (LR+ 6-33), and a description of “tearing pain” (LR+ 10.8). But many other signs are sensitive but not that specific, or not even that sensitive (and ‘tearing pain’ is only reported 39% of the time, with focal neuro deficits found just 17% of the time). Chest Xrays have a LR+ of around 2, and are only 60-80% sensitive (though a completely normal CXR — normal width mediastinum and normal aortic knob, have a LR- 0f 0.3).

What’s really needed to diagnose AAD is not-so-standard imaging, like CT, TEE, or MRI (all 98% sensitive or greater, with specificies in the high 90’s as well). So there’s an opportunity for a lab test to step in and help us guide decisionmaking. And d-dimer, a substance released by fibrinolytic activity unleashed when the extrinsic pathway is activated in dissection, in theory could work well as a marker.

The aim of the Ohlmann paper was to determine (in a larger cohort than had been tried previously) the value of d-dimer measurements in diagnosis and prognosis of patients suspected of AAD. How did they try to show d-dimer is diagnostic and prognostic? They used a single (French) center retrospective case-control trial, looking at 6 years of patients, spanning just 16,529 records, and identified 94 consecutive patients with AAD as diagnosed by imaging (TEE, CT, MRI and/or autopsy) who also had dimers drawn. Now, it was very confusing to us who these 16,529 patients were, and how “94 consecutive” AAD patients were identified. How many patients came through the door in those 6 years? With chest pain? With AAD, but not dimers? What does “consecutive” even mean in this context? It’s not revealed, and it’s a serious shortcoming.

Then, these 94 patients were matched to a group of controls. How they were matched is also not revealed — it seems they were age-matched because their ages are similar. But the groups had significantly dissimilar rates of hypertension, back pain, “intensive pain” (21% vs. 4%), not to mention neuro deficits, pulse deficits, and syncope. What’s more, the control group wasn’t just undifferentiated, vague chest pain — but a group of patients with dizzyingly high levels of sickness (STEMI 17%, PE 13%, NSTEMI 13%, pericarditic 7%, aortic aneurysm without dissection 16%, PE 13%, and just 6% that sounded like GERD and just 18% with chest wall pain).

On average, D dimer levels were measured 1.2 days after symptoms started, +/- 2.5 days, but at least before surgery. It’s not at all clear if the dimer came first or influenced decision making. Most of their AAD was diagnosed by CT and/or TEE, with a few getting the AAD diagnosis by MRI or angio or autopsy. Most were standford A, and a few had intramural hematomas.

99% of patients with AAD had elevated D-dimer. Those with intramural hematomas had lower d-dimers, and the one false negative guy had a intramural hematoma without intimal flap. D-dimers were higher in Stanford A vs. B, and higher in patients with classifications of Debakey I > II > III.

66% of controls also had elevated D-dimer, though it was significantly lower. D-Dimer was highly predictive of AAD, with 99% sensitivity and 34% specificity. LR+ was 1.5 with a high d-dimer, and LR- was 0.03… D-dimer was not correlated with CRP or WBC or troponin I.

Several members of our group noted that, if d-dimer as a test works, higher d-dimers should correlate with bigger dissections, and that indeed seems to be the case. Also, higher d-dimer levels correlated with death.

All told, it’s a good promising study design, damaged by poor reporting of the methods. They still manage to suggest that d-dimer can be useful when it comes back negative — but what it means when it’s positive is far from clear.

As one attending in our grouip reflected, our approach to d-dimer in AAD now may mirror our approach to PE from ten years ago. That is to say, dimers will be ordered willy-nilly until some criteria are developed to proscribe their use, then eventually, we’ll develop criteria to rule out AAD without labs altogether, a la PERC. Until then, I think it’s likely d-dimers will be ordered on low-risk patients, some will come back falsely positive, and we’re going to need a lot of imaging ruling out a rare, rare entity. In the meantime, I’m going to refine my pretest probability with some old-fashioned history and physical exam findings, plus the ever-essential vigilance for rare diagnoses.

Posted on Friday, August 8th, 2008 at 12:47 am by Nick. Filed under Risk Stratification, Journal Club, Blog.
You may post a comment.