Outcomes of Patients Presenting With Acute Type A Aortic Dissection in the Setting of Prior Cardiac Surgery
An Analysis From the International Registry of Acute Aortic Dissection
Background—Prior cardiac surgery (PCS) can complicate the presentation and management of patients with type A acute aortic dissection (TAAAD). This report from the International Registry of Acute Aortic Dissection examines this hypothesis.
Methods and Results—A total of 352 of 2196 patients with TAAAD (16%) enrolled in the International Registry of Acute Aortic Dissection had cardiac surgery before dissection, including coronary artery bypass grafting (34%), aortic or mitral valve surgery (36%), aortic surgery (42%), and other cardiac surgery (16%). Those with PCS were older, had a higher frequency of diabetes mellitus, hypertension, and atherosclerosis, and presented later from symptom onset to hospital presentation and diagnosis (all P<0.05). In-hospital mortality was significantly higher for PCS patients (34% versus 23%; P<0.001). Five-year mortality was independently predicted by PCS (hazard ratio [HR], 2.04; 95% confidence interval [CI], 1.05–3.95), age >70 years (HR, 2.65; 95% CI, 1.40–5.05), medical management (HR, 5.10; 95% CI, 2.43–10.71), distal communication (HR, 2.64; 95% CI, 1.35–5.14), and coma (HR, 9.50; 95% CI, 2.05–44.05). Among patients with PCS, in-hospital (43% medical versus 30% surgical; P=0.033) and intermediate-term mortality was higher in patients with medical versus surgical management. Propensity-matched analysis revealed significant increase in mortality with medical management, but not with PCS.
Conclusions—PCS delays presentation, diagnosis, and treatment of TAAAD and is an important adverse risk factor for early and intermediate-term mortality. This effect may be because of increased medical management in this patient population.
Presentation with type A acute aortic dissection (TAAAD) is associated with high mortality and a 5-year survival of 32%, with operative therapy indicated to improve survival.1,2 Prior cardiac surgery (PCS) may impact the presentation and outcomes of patients with TAAAD, and the ideal management of these patients is not clear. The reported operative mortality for repair of TAAAD occurring in patients with PCS varies from 5% to 54%.3,4 Some have argued a history of PCS should favor a nonoperative approach.5–7 Several justifications are offered for this, including protection from aortic rupture and aortic insufficiency conferred by adhesions from previous surgery and previous aortic valve replacement, respectively.8
This study from the International Registry of Acute Aortic Dissection (IRAD) aims to determine the impact of PCS on the presentation and outcomes of patients with TAAAD. We hypothesize that PCS increases the risk of morbidity and mortality in patients who develop type A dissection, and that operative therapy should remain the preferred treatment modality in appropriately selected patients.
The IRAD registry is composed of 31 centers in 11 countries. Participating centers report consecutive cases of acute dissection to the IRAD Coordinating Center at the University of Michigan, which performs internal checks for validity. The institutional review board at each center approved data collection. Further details of the registry have been published previously.2
Patients presenting with TAAAD to an IRAD site formed the study cohort. Those with no prior history of PCS were compared with those who had PCS. The primary outcome measure was in-hospital mortality. Secondary outcome measures included early morbidity and intermediate-term mortality.
A total of 2196 patients presented with TAAAD from January 1996 to March 2012. Type A dissection was defined by the Stanford classification. A total of 352 of these patients had PCS, including aortic surgery (149, 42%), aortic or mitral valve replacement (128, 36%), coronary artery bypass grafting (118, 34%), and other cardiac surgery (56, 16%). Of these 352 patients, 266 (76%) were treated operatively, 74 (21%) were treated medically, and 12 (3%) were treated with endovascular or hybrid therapy. Medical management referred to patients managed nonoperatively, generally with blood pressure optimization and close hemodynamic monitoring, at the discretion of the treating hospital.
Primary analysis compared patients with TAAAD with and without PCS. Categorical variables were compared using χ2 analysis or Fisher exact test where appropriate. Continuous variables were analyzed with 2-sample t tests or Wilcoxon rank-sum tests. Multiple logistic regression analysis was performed to create an in-hospital mortality model, and a Cox proportional hazards regression analysis was performed to create a 5-year mortality model. Variables introduced to the models included those from univariate analysis with P<0.20. An additional model was created that was propensity-adjusted for the likelihood of receiving medical management. Univariate factors associated with medical management (P<0.20) were introduced to the model, and a backward stepwise regression technique was used to develop the variables to be included in the final model. The resulting regression equation was then used to calculate a conditional probability of medical management, and this was used as a variable to propensity adjust the mortality models. All variables were checked for proportionality with standard plots of cumulative hazard versus time and log minus log versus time. Kaplan–Meier analysis generated 5-year survival curves, and the log-rank test was used to compare survival between groups. All statistics were performed with SPSS version 20 (IBM Corporation). Significance was defined as a 2-sided P value <0.05.
Baseline characteristics and clinical presentation of patients with TAAAD with and without PCS are shown in Table 1. Patients with PCS were more likely to be older, men, and have a history of hypertension, atherosclerosis, diabetes mellitus, and Marfan syndrome. They were also more likely to have a previous aortic dissection or an iatrogenic dissection. Although the location of prior dissection is unknown (type A versus type B), all dissections in the current study were new, as patients with extension of a previous dissection are not included in IRAD. Iatrogenic dissections in PCS patients were caused by cardiac surgery in 75% of patients, cardiac catheterization in 21%, and other causes in 4%. Patients without a history of PCS were more likely to be managed surgically for their dissection than patients with PCS (87% versus 76%; P<0.001). Operative details differed between groups and are shown in Table 1. Cerebral perfusion strategies, including use of antegrade (52% PCS versus 56% non-PCS) or retrograde perfusion (23% PCS versus 15% non-PCS), were similar (P=0.51).
Early outcomes are shown in Table 2. PCS patients surprisingly had a lower rate of stroke but significantly higher in-hospital mortality. When subjected to multivariable logistic regression, age >70 years, PCS, and medical management were identified as independent predictors of in-hospital mortality (all P<0.05; Table 3). Iatrogenic dissection was not associated with worse outcomes in PCS patients.
Intermediate-term survival by Kaplan–Meier analysis is shown in Figure 1. The 5-year actuarial survival was 82% (54% with PCS, 85% without PCS) with a median follow-up of 24 months (Q1–Q3, 12–48 months). Log-rank analysis revealed significant reduction in survival during the 5-year follow-up period in PCS patients (P<0.001). By multivariable Cox regression, independent predictors of intermediate-term mortality from initial hospitalization included age >70 years, PCS, and medical management, distal communication of the dissection flap seen on diagnostic imaging, and coma (all P<0.05; Table 3).
Medical Versus Operative Therapy in Patients With PCS
Given previous reports suggesting suitability of a nonoperative approach for patients with PCS, we compared patients who underwent an operation with those treated medically. Table 4 shows the baseline characteristics, clinical presentation, and early outcomes of the 266 PCS operatively managed patients versus the 74 PCS patients treated medically. Patients undergoing an operation were younger, had a lower rate of atherosclerosis, and also presented more frequently with chest pain and pericardial effusion. Early outcomes favored an operative approach, with this group of patients having lower rates of hypotension, extension of dissection, spinal cord ischemia, and in-hospital mortality. A Kaplan–Meier analysis of intermediate-term survival is shown in Figure 2, with a significant reduction in survival in medically treated patients (log-rank P<0.001).
To correct for the difference in underlying comorbidities between the medical and operative therapy groups, we performed a propensity-matched analysis, incorporating the likelihood of medical therapy into models of in-hospital and intermediate-term mortality. These models are shown in Table 5. The model of in-hospital mortality has a C-statistic of 0.859 and a Hosmer–Lemeshow P=0.289. In this propensity-matched model, PCS is no longer significantly associated with increased hospital mortality.
Aortic dissection occurring in the setting of PCS may have a different natural history than that occurring de novo. In their series, Gillinov et al8 found that only 18% presented with aortic rupture and only 6% had hemodynamic compromise from their rupture. Others have also advocated a conservative approach, citing excellent late survival (80%–100% at 1.5–6.5 years) with primary medical management.5–7 However, these studies had very small sample sizes and included patients who were clinically stable at the time of presentation. The presence of adhesions and their presumed protective effect against rupture and tamponade has been suggested as a potential reason for improved survival with medical therapy. However, adhesions would not likely protect against other modes of mortality from untreated type A dissection, including aortic insufficiency, coronary artery occlusion, or fistula formation into other cardiac chambers.
TAAAD repair in the reoperative setting is a formidable procedure, associated with significant morbidity and mortality. Several studies have shown that patients with PCS have higher in-hospital mortality after aortic repair.9–12 Centofanti et al9 created a multivariate prediction model for 30-day mortality after surgery for aortic dissection. In addition to PCS, independent predictors of 30-day mortality included shock, coma, advanced age, and renal failure. They argue for conservative management in patients who have a higher mortality rate predicted by their model than that which would be expected from primary medical management alone. A more recent study from the Netherlands looked at predictors of mortality in patients with type A dissection. Their study cohort of 232 patients included 24 with PCS. In contrast to other reports, they found no correlation between PCS or prior aortic surgery and early or late mortality.13
In the current study, patients with PCS had a lower rate of postoperative stroke than those without prior surgery. This finding on univariate analysis was unexpected, and the explanation for this is not immediately clear. Estrera et al10 reported that patients with PCS had a higher rate of stroke.
A previous IRAD report on 100 patients with TAAAD in the setting of PCS was published in 2004 and described only presenting characteristics and in-hospital outcomes. Our findings on patient characteristics and clinical presentation were similar to those from 2004. However, no difference in in-hospital mortality was identified in that smaller group.14 This current study expands on our previous work, and the addition of more patients reveals significantly different outcomes in patients with PCS. More importantly, the current study describes a much larger cohort and also includes intermediate-term mortality stratified by a history of PCS and treatment type.
Our study demonstrates the adverse impact of PCS on early morbidity and mortality in patients who present with TAAAD. Patients with PCS not only were less likely to have the classic symptoms of dissection but also presented more frequently with a delay from symptom onset to hospital presentation, diagnosis, and aortic repair. This study also identified a clear difference in intermediate-term survival between type A dissection patients with or without PCS. This is a unique finding in the literature, as even the studies that reported a difference in early mortality were unable to demonstrate long-term survival differences.10,12
Our study supports the protective effect of PCS on rupture and tamponade, given that these patients had a lower rate of pericardial effusion and hypotension. However, the conclusion drawn by some, that this justifies nonoperative management in these patients, is not supported by our data. Recognizing the increased mortality associated with type A dissection in the setting of PCS, we also compared medical therapy with primary operative therapy in patients with PCS, to our knowledge the largest such analysis in the literature. Medically managed PCS patients had an in-hospital mortality rate of 43%, compared with 30% in the surgically managed group. Furthermore, this mortality difference persists after discharge. Medically managed patients who survive to hospital discharge have significantly worse intermediate-term mortality than those patients who had operative dissection repair.
We performed a propensity-matched analysis, which incorporated the likelihood of medical management, as opposed to operative management. In this model, independent predictors of in-hospital mortality include medical management, as well as several in-hospital complications (acute renal failure, mesenteric ischemia, hypotension, and coma). PCS was not a predictor of in-hospital or intermediate-term mortality in this model. Given the significantly higher percentage of PCS patients who are managed medically, this suggests that the increased mortality seen in this patient population might actually be driven by the decreased operative intervention for TAAAD. These results justify recommendations to proceed with aortic repair of type A dissection in all patients with acceptable comorbidities, regardless of whether the patient previously had cardiac surgery.
There are several limitations to this study. First, the data are collected both retrospectively and prospectively, and the potential exists for reporting bias. Second, the retrievable data are limited to that obtained from the IRAD data form. Specifically, we have little to no information on details and timing of the prior operations (including location of prior aortic surgery) performed, and not all patients had data reported for all variables. In addition, we do not have details on the timing and location of any prior aortic dissections in these patients. Finally, as with any retrospective aortic dissection study, it is impossible to determine the number of patients who developed an acute dissection and died before IRAD center arrival. Thus, we are unable to determine the true mortality rates for the conditions described. Despite these limitations, this study provides unique information from a large multicenter registry with both early and intermediate-term data.
Patients presenting with TAAAD in the setting of PCS more frequently have atypical symptoms and are diagnosed later than those without PCS. Operative therapy in this setting is associated with increased in-hospital and intermediate-term mortality when compared with those without prior history of cardiac surgery. However, medical therapy alone is associated with a higher in-hospital and intermediate-term mortality and seems to be the primary determinant of mortality in patients with PCS, thus supporting an operative approach for TAAAD in the setting of a prior history of cardiac surgery.
Sources of Funding
The International Registry of Acute Aortic Dissection Registry is funded through grants provided by W.L. Gore and Associates, Inc, the Varbedian Aortic Research Fund, the Hewlett Foundation, the Mardigian Foundation, University of Michigan Faculty Group Practice, and Terumo.
Dr Eagle received research grant from W.L. Gore; Dr Fillinger received research grant from and is a consultant for W.L. Gore, Cook. The other authors report no conflicts.
- © 2013 American Heart Association, Inc.
- Hagan PG,
- Nienaber CA,
- Isselbacher EM,
- Bruckman D,
- Karavite DJ,
- Russman PL,
- Evangelista A,
- Fattori R,
- Suzuki T,
- Oh JK,
- Moore AG,
- Malouf JF,
- Pape LA,
- Gaca C,
- Sechtem U,
- Lenferink S,
- Deutsch HJ,
- Diedrichs H,
- Marcos y Robles J,
- Llovet A,
- Gilon D,
- Das SK,
- Armstrong WF,
- Deeb GM,
- Eagle KA
- Stanger O,
- Oberwalder P,
- Dacar D,
- Knez I,
- Rigler B
- Modi A,
- Vohra HA,
- Kaarne M,
- Haw MP,
- Barlow CW,
- Ohri SK,
- Livesey SA,
- Tsang GM
- Bekkers JA,
- Raap GB,
- Takkenberg JJ,
- Bogers AJ
- Collins JS,
- Evangelista A,
- Nienaber CA,
- Bossone E,
- Fang J,
- Cooper JV,
- Smith DE,
- O’Gara PT,
- Myrmel T,
- Gilon D,
- Isselbacher EM,
- Penn M,
- Pape LA,
- Eagle KA,
- Mehta RH