D-dimer and the Aortic Dissection Detection Risk Score
Washington University Emergency Medicine Journal Club– November 18th, 2021
You are moonlighting in a local community ED one afternoon—making serious bank and contemplating what to spend on that money on—when you encounter a forty-year old male patient with chest pain. He has a history of hypertension, for which he takes lisinopril, and chronic kidney disease. He was watching TV earlier in the day, eating a chili dog, when he developed a fairly abrupt onset of dull pain in the center of his chest.
As you continue your HPI, you learn that the pain does, in fact, radiate into his back, but is not ripping or tearing. He has no neurologic deficits, has symmetric blood pressures in both arms, and has stable vitals with an elevated blood pressure (BP 180/95, HR 79, RR 12, SpO2 98%). You order and ECG, CXR, cardiac enzymes, a BMP, and CBC, all of which come back normal aside from a creatinine of 2.1 (which is unchanged compared to prior records).
You are still considering an acute aortic dissection as part of your differential, given the radiation of pain to the back, but would prefer not to expose your patient to the risks of radiation and the theoretical risk of contrast unless absolutely necessary. Your suspicion is fairly low. You recently heard a talk on EM:RAP regarding the use of D-dimer and something called an Aortic Dissection Detection Score, but haven’t seen in evidence support to support such a work-up. You find a review of this topic in the Journal Club Archives (October 2015), but the results were felt to be inconclusive at that time. You wonder what additional evidence has been published in the interim.
In the end, you get a CT scan, which is negative for dissection, and admit the patient for blood pressure control and a possible stress test. When you get off work, you decide to search the evidence and see what you can find.
Population: Adult ED patients with suspected aortic dissection
Intervention: D-dimer in conjunction with the Aortic Dissection Detection Score
Comparison: CT aortic antiogram, MRI
Outcome: Diagnostic accuracy (sensitivity, specificity, likelihood ratios)
PubMed was searched using the terms aortic dissection detection AND “D-dimer”
resulting in 47 citations (https://tinyurl.com/2p8arGj8). Among these, the four most
relevant articles were selected. These included two retrospective studies, one
prospective observational trial, and one systematic review and meta-analysis.
Article 1: Bima P, Pivetta E, Nazerian P, Toyofuku M, Gorla R, Bossone E, Erbel R, Lupia E, Morello F. Systematic Review of Aortic Dissection Detection Risk Score Plus D-dimer for Diagnostic Rule-out Of Suspected Acute Aortic Syndromes. Acad Emerg Med. 2020 Oct;27(10):1013-1027. doi: 10.1111/acem.13969. Epub 2020 Apr 21. PMID: 32187432. Answer Key
Article 2: Nazerian P, Mueller C, Soeiro AM, Leidel BA, Salvadeo SAT, Giachino F, Vanni S, Grimm K, Oliveira MT Jr, Pivetta E, Lupia E, Grifoni S, Morello F; ADvISED Investigators. Diagnostic Accuracy of the Aortic Dissection Detection Risk Score Plus D-Dimer for Acute Aortic Syndromes: The ADvISED Prospective Multicenter Study. Circulation. 2018 Jan 16;137(3):250-258. doi: 10.1161/CIRCULATIONAHA.117.029457. Epub 2017 Oct 13. PMID: 29030346. Answer Key
Article 3: Nazerian P, Morello F, Vanni S, Bono A, Castelli M, Forno D, Gigli C, Soardo F, Carbone F, Lupia E, Grifoni S. Combined use of aortic dissection detection risk score and D-dimer in the diagnostic workup of suspected acute aortic dissection. Int J Cardiol. 2014 Jul 15;175(1):78-82. doi: 10.1016/j.ijcard.2014.04.257. Epub 2014 May 2. PMID: 24838058. Answer Key
Article 4: Gorla R, Erbel R, Kahlert P, Tsagakis K, Jakob H, Mahabadi AA, Schlosser T, Eggebrecht H, Bossone E, Jánosi RA. Accuracy of a diagnostic strategy combining aortic dissection detection risk score and D-dimer levels in patients with suspected acute aortic syndrome. Eur Heart J Acute Cardiovasc Care. 2017 Aug;6(5):371-378. doi: 10.1177/2048872615594497. Epub 2015 Jul 16. PMID: 26185259. Answer Key
Acute aortic syndromes (AASs)—including aortic dissection, intramural aortic hematoma, and penetrating aortic ulcer—while rare, are associated with poor outcomes. Approximately 17% of patients with AAS will die prior to reaching the hospital, with an additional 21% dying after arrival (Melvinsdottir 2016). Delayed diagnosis can be catastrophic, as mortality increases by 1-2% with each hour following symptom onset (Harris 2011). Unfortunately, the typical signs and symptoms of AAS are neither specific nor sensitive for the disease (Klompas 2002). While computed tomographic angiography (CTA) is highly accurate in the diagnosis of AAS, its use is associated with significant cost and the risks of radiation and contrast exposure (Urbania 2009). Magnetic resonance angiography (MRA) and trans-esophageal echocardiography (TEE) are also diagnostic options, but again incur significant cost with the added risks of transport away from the ED in the case of MRA and procedural complications for TEE (Sentz 2015).
In 2010, the American Heart Association and American College of Cardiology released guidelines which included a list of high risk features for acute aortic dissection. These were later used to form the Aortic Dissection Detection risk score (ADD-RS) which was shown to be helpful in risk stratifying patients in the International Registry of Acute Aortic Dissection into low (4.3%), intermediate, and high risk groups (Rogers 2011). When used in conjunction with a negative D-dimer,—whose negative likelihood ratio was 0.05 in a previously published meta-analysis,—a theoretical low-risk patient with an ADD score of 0 and a negative D-dimer would have a post-test probability of disease of 0.2%, and hence may not need to undergo imaging.
There have since been additional studies evaluating an algorithm combing a low-risk ADD-RS with D-dimer to rule out AAS. A retrospective study out of Germany (Gorla 2015) found that among 376 patients being evaluated for aortic dissection, 127 (34%) had an ADD score of 0 and a negative D-dimer. None of these patients were found to have an aortic dissection (sensitivity 100%, specificity 67.5%, negative predictive value (NPV) 100%, negative LR 0). Among 57 patients with an ADD score of 1 and a negative D-dimer, two were found to have an AAS. The overall NPV of an ADD score of ≤ 1 and a negative D-dimer was 98.9% (95% CI 95.9 to 99.7%).
In a similar study from Italy, a prospectively collected registry was used to retrospectively evaluate the ADD score combined with D-dimer testing (Nazerian 2014). Among 1035 patients being evaluated for aortic dissection with a D-dimer level available, 92 (8.9%) had both an ADD score of 0 and a negative D-dimer, and none of these patients were found to have an aortic dissection (NPV 100%, negative LR 0). Among 152 patients with an ADD score of 1 and a negative D-dimer, only 2 (0.2%) were found to have an aortic dissection. The overall NPV of an ADD score of ≤ 1 and a negative D-dimer was 99.2% (95% CI 96.8 to 99.8%).
While these studies offer promising results, both are at high risk of bias due to their retrospective nature. The ADD-RS was calculated retrospectively in both studies, forcing the authors to impute potentially key data when they were missing. Additionally, both studies included only those patients with a D-dimer result reported rather than all patients in whom aortic dissection was clinically suspected (selection bias).
There has been a single prospective study evaluating the diagnostic accuracy of an ADD-RS/D-dimer algorithm (Nazerian 2018). The ADvISED trial was an international, multicenter trial that prospectively enrolled 1850 patients with suspected AAS. A negative D-dimer alone had a negative LR for AAS of 0.05 (95% CI 0.03 to 0.1), similar to prior reports. Out of 294 patients with an ADD-RS of 0 and negative D-dimer, only one was diagnosed with an AAS (NPV 99.7%, 95% CI 98.1 to 100%). Among 924 patients with an ADD-RS ≤ 1 and a negative D-dimer, 3 cases of AAS were observed, resulting in a NPV of 99.7% (95% CI 99.1 to 99.9%). The disease prevalence of 0.3% in this low-risk group is clearly below the previously reported test threshold for CTA for AAS of 3% (Sarasin 1996). While this calculation is based on outdated data, it seem unlikely that an updated test threshold would fall below 0.3%.
A systematic review and meta-analysis on this topic pooled results from these three studies in addition to a fourth retrospective study (Bima 2020). ADD-RS = 0 and D-dimer < 500 had a pooled sensitivity of 99.9% (95% CI 99.3% to 100%), a negative LR of 0.032 (95% CI 0 to 0.086), and a failure rate of 0.1% (95% CI 0% to 0.3%). An ADD-RS = 0 combined with a D-dimer < 500 was 99.9% sensitive with a negative LR of 0.032 (95% CI 0 to 0.086), and a failure rate of 0.1% (95% CI 0% to 0.3%). Using an ADD-RS ≤ 1 and D-dimer < 500, the sensitivity dropped slightly to 98.9%, with a negative LR of 0.025 (95% CI 0.001 to 0.049) and a failure rate 0.6% (95% CI 0.2% to 0.9%).
While the use an ADD-RS/D-dimer algorithm appears promising in the evaluation of AAS, the bulk of this data is derived from lower quality retrospective studies at high risk of selection bias, inaccuracies due to missing data, and differential verification bias. Based on the more methodologically rigorous ADvISED study, it would be reasonable to use a low-risk ADD-RS score in combination with a negative D-dimer to rule out AAS. As this was a single study conducted outside of the US, and given the high stakes of a missed diagnosis, it would also be reasonable to await further validation of this diagnostic protocol prior to use. Given variability in risk tolerance among emergency physicians, individual clinicians will have to decide what level of evidence is required for integration of this clinical decision rule into practice.