Modifiable Risk Factors for Early Mortality in Low-Risk Penn Class Aa Acute Type A Aortic Dissection Patients – A Descriptive Study

Abstract Background: Acute Type A aortic dissection (ATAAD) without end-organ or generalized ischemia is Penn class Aa and considered low-risk. Nevertheless, surgical mortality can be considerable in this subgroup and may be related to modifiable factors. The objective of this study was to analyze 30-day mortality among ATAAD Penn class Aa patients with special reference to modifiable perioperative factors. Methods: Among all patients operated for ATAAD from 1990 to 2010, all Penn class Aa patients dying within 30 days were included in a retrospective descriptive study. Pre- and intraoperative variables related to 30-day mortality were retrieved from medical records and analyzed according to avoidable or modifiable errors such as initial misdiagnosis, preoperative delay, adverse events, and forced and unforced additional procedures. Results: Overall 30-day mortality was 13% (31/235). Intraoperative death occurred in 32% (10/31) of patients. Among patients not dying intraoperatively, stroke was the most common complication (48%) and cause of death overall, followed by reoperation for bleeding (33%), respiratory failure (24%), and renal failure (14%). Preoperative errors were detected in 48% of patients; one-third had initial misdiagnosis and/or diagnostic delay ≥ 24 hours. Intraoperative error(s) was noted in 74% of patients, mainly involving adverse event(s), forced additional procedures, and assisted bleeding control, with each affecting approximately 45% of patients. Conclusion: Modifiable errors in pre- and intraoperative management are not uncommon among Penn class Aa patients and may contribute to 30-day mortality. Efforts should be made to mitigate the effects of modifiable factors to reduce early mortality in low-risk ATAAD.


Introduction
The Penn classification for acute Type A aortic dissection (ATAAD) is a simple, reproducible, and validated means of stratifying the risk of early (30-day) mortality due to ischemic complications [1][2][3][4]. In the original Philadelphia study, 30-day mortality was 3.1% in Penn class Aa (i.e., uncomplicated ATAAD) patients versus 17.6-40.0% in Penn class non-Aa patients [1]. This finding demonstrates that Penn class Aa patients can be considered relatively "low-risk", although 30day mortality in this group is more often reported in the 10-15% range [2,4]. However, why Penn class Aa patients die is unclear, as deranged preoperative physiology secondary to localized or generalized ischemia is, by definition, not present. Therefore, the objective of this study was to analyze the cause of early death of Penn class Aa ATAAD patients. More specifically, we investigated putatively influential pre-and intraoperative factors (e.g., modifiable errors) that could contribute to early mortality in the absence of increased risk for cardiac and multiorgan failure related to preoperative ischemia. Identifying modifiable factors could help improve early outcomes of both Penn class Aa and non--Aa ATAAD patients.

Patient Population
The underlying patient cohort from which the present study population was obtained was previously described in detail [2] and consisted of 360 consecutive patients undergoing surgery for ATAAD from 1990 to 2009. This study was extended with operations performed in 2010 to include an additional 31 patients (n = 391). After re-examination of all case notes, 72 (18%) early (i.e., in-hospital) deaths were identified. To conform to the original Penn classification cohort definitions, nine patients were excluded (three surviving more than 30 days, six operated with-out hypothermic circulatory arrest) to form the final study population of 31 patients ( Figure 1). Thus, the study population comprised only Penn class Aa patients dying within 30 days. Demographic and basic clinical characteristics of patients in the study group (died within 30 days, n = 31) and, for comparison, operative (>30 days) survivors in Penn class Aa (n = 189) are shown in Table 1.

Study Variables and Definitions
Age, gender, weight, height, date of operation, and date of death were collected for all patients. Family history was defined as a known occurrence of an aortic condition in a first-degree relative. Chronic obstructive pulmonary disease, diabetes, hypertension, previous cerebrovascular accident, Marfan syndrome, and shock were defined in accordance with the Society of Thoracic Surgeon's database criteria. The extent of dissection was categorized as DeBakey I (entire aorta) or II (ascending aorta, including but not distal to the arch) [5]. Penn class was defined as Aa with no clinical ischemic complications (localized or generalized), Ab with localized ischemia of one or more organ systems (brain, spinal cord, kidneys, abdominal viscera, or extremities), Ac with generalized ischemia (also including myocardial ischemia), and Abc with the combination of Ab and Ac [1]. Of note, radiological findings of malperfusion without clinical correlation were considered Penn class Aa (e.g., hemodynamically stable patients with tamponade).
Putatively modifiable pre-and intraoperative variables were examined in detail. Preoperative variables included initial misdiagnosis (explicitly affecting diagnostic or therapeutic procedures), time elapsed from symptoms to surgical treatment (any delay categorized as ≥ 24 or 48 hours); use of coronary angiography for suspicion of acute coronary syndrome (ACS) and not as part of a planned preoperative work-up, and medical treatment (thrombolysis and/or antithrombotic or antiplatelet regimens) for ACS. Intraoperative variables included errors related to adverse events, additional procedures, residual dissection, severe hemorrhage, and surgeon experience ( Table  2). Assisted bleeding control was defined as pharmacological or surgical methods to achieve hemostasis that went beyond first-line management, such as mediastinal tamponade with delayed sternal closure, cabrol shunt, or NovoSeven®.

Surgical and Perioperative Procedures
All operations were performed through median sternotomy on cardiopulmonary bypass using hypothermic circulatory arrest. Utilization of ante-or retrograde cerebral perfusion for cerebral protection was at the discretion of the surgeon. Sites of arterial and venous cannulation varied but typically included the femoral artery or right axillary artery and the femoral vein or right atrium, respectively. Myocardial protection was typically provided by a combination of cold blood ante-(coronary osteal) and retrograde cardioplegic arrest. The most common surgical procedure was a supracoronary graft replacement with an open distal anastomosis often in the shape of a hemiarch replacement, with resuspension of the aortic valve and Teflon felt suture line reinforcement. Aortic root replacement and total arch replacement were used selectively, and additional surgical procedures, such as separate valve repair/replacement or coronary artery bypass grafting, were performed after individual assessment. Main intraoperative procedures are summarized in Table 3. BioGlue® was not used, and gelatin-resorcinol-formaldehyde glue was discontinued around 2004.

Statistical Analysis
Data were summarized using descriptive statistics (i.e., frequency with percentage, median with interquartile range). Analyses were performed using Stata 13 (Stata Corp LP, College Station, TX).
In the preoperative phase, initial misdiagnosis occurred in over one-third of patients. A suspicion of ACS was most common, followed by, in equal numbers, suspicion of pulmonary embolism, acute abdomen, and ruptured abdominal aortic aneurysm. The delay from first documented symptoms to surgical treatment was > 24 hours for 32% of patients and > 48 hours for 26% of patients (Table 4). Preoperative coronary angiography (for suspected ACS) was not performed, but inadvertent pharmacological treatment for ACS did occur. and could include very potent pharmacotherapeutics such as thrombolysis for suspected ACS. Delay of surgical treatment ≥ 48 hours was noted in 26% of cases. Hemodynamic instability and ECG alterations are likely to occur more often in Penn class non-Aa patients, whereas Penn class Aa patients may be completely stable without presentation of alarming symptoms, contributing to broadened and/or delayed diagnostic work-up. Whether increasingly sensitive ACS diagnostics combined with earlier, often pre-hospital, initiation of ACS treatment will affect this scenario remains to be determined. In this study, management delay had no apparent effect on outcomes, although most physicians and surgeons would agree that such delays should always be avoided and that stable Penn class Aa patients can rapidly deteriorate. Adverse intraoperative events occurred frequently, and, as a result, forced additional procedures such as coronary artery bypass grafting and graft exchange were also common. At least four adverse events appeared as surgical technical failures. Re-applying the aortic cross-clamp to address an intraoperative prob-Intraoperative modifiable errors occurred in 74% of cases (Table 5). Severe hemorrhage, adverse events, re-clamping of the aorta, and forced additional procedures were relatively common, with 41% of cases involving three or more such intraoperative errors. The frequency and nature of adverse events and forced additional procedures are shown in Table 6. High-volume surgeons performed approximately half of the operations in both groups.

Discussion
In Penn class Aa ATAAD patients, the absence of preoperative deranged physiology or end-organ ischemia entail lower risk of cardiac and multiorgan failure as well as lower risk of critical end-organ dysfunction, which are conditions associated with early mortality [6].

Situation Description
Adverse events -Coronary artery destruction and disintegration (n = 4) -Aortic rupture (n = 3) -Sudden tamponade (n = 2) -Ventricular fibrillation prior to cardiopulmonary bypass -Venous air embolism from Cellsaver circuit -Compromised flow and arterial recannulation to contralateral femoral artery -Femoral venous cannula atrial septal perforation -Liver injury from rapid sternotomy lem is likely a strong marker of significant technical problems that were not resolved by other means. Restarting cardiopulmonary bypass, on the other hand, may be forced by different conditions such as hemorrhage, respiratory failure, cardiac failure, volume overload, and protamine reaction.
Severe bleeding was also common, both in terms of assisted bleeding control and re-exploration. Blood product usage data were not available for analysis. Blood-conserving strategies vary, but a more liberal or even routine use of recombinant factor VIIa (NovoSeven®), for example, is becoming more popular. In this context, we used the term assisted bleeding control to describe when extraordinary measures were needed due to bleeding, a general situation applicable regardless of local treatment protocols. However, bleeding as cause of death was not dominating; rather, brain damage prevailed as the most common cause of death. The study design did not allow further analysis of whether strokes were predominantly of an embolic or hypoperfusion nature; as etiology varies, so does their potential modifiability.
Despite increased occurrence of delayed surgical treatment, adverse events, additional procedures, and intraoperative failures, intraoperative death was not excessively common. This could be related to the better preserved preoperative physiological status of Penn class Aa patients as compared with their class non-Aa counterparts. High-volume surgeons operated on patients in both groups in equal proportions, and surgeon experience was not statistically related to adverse events, additional procedures, or assisted bleeding control. In this retrospective study based on medical records, it was not possible to discern the degree to which individual deaths could have been avoided or should be regarded as unnecessary, but it must be argued that several variables related to early death in Penn class Aa ATAAD patients were modifiable and could be improved.
Recent reports suggest that the concept of "lowrisk" ATAAD is reasonable. The operative risk of the typical (i.e., 62-year-old male without significant comorbidity) Penn class Aa patient according to Euro-Score II is around 2.0%. In the original Philadelphia study, 30-day mortality was 3.1% in Penn class Aa [1] and equally low (3%) in a Japanese experience [3]. Some centers of excellence even report overall early mortality of 3.3-5.3% in smaller (n = 72-107) case se-ries [7][8][9]. In comparison, larger single-or multicenter studies typically report 17-26% early mortality [2,4,[10][11][12][13]. Notably, these estimates are more similar to those for high-risk groups corresponding to Penn class non-Aa, ranging from 15% to 45% [1][2]4]. Thus, it is important to improve outcomes of Penn class Aa ATAAD and to further analyze such low-risk deaths. The seminal Cambridge FIASCO study [14] (n = 4294) analyzed early death in general low-risk cardiac operations and found that technical errors were probable in 7/16 instances of preventable death and concluded that quality improvement programs may further reduce such mortality, as evidenced by the subsequent FIASCO II study [15] showing similar crude mortality but a lower proportion of preventable deaths. The findings of the present study also indicate that preand intraoperative avoidable errors contribute to early mortality in the otherwise low-risk Penn class Aa group and that these factors are potentially modifiable by improved diagnostic procedures, improved primary surgical procedures including complete resection of the primary entry, and meticulous attention to surgical technical detail.
Recently, using a similar study design, Buonocore et al. investigated surgical factors related to their 23% 30-day mortality in 111 consecutive Penn class Aa patients [16]. They found an association between " non-aortic surgeon" and early mortality (odds ratio 6.40 in multivariable analysis) and noted that aortic surgeons used more extensive procedures. However, they did not investigate intraoperative details. Thus, the present study findings may point to the occurrence of intraoperative adverse events, rather than operator experience per se, as related to early outcome.
In summary, we found that death of Penn class Aa ATAAD patients was often related to intraoperative avoidable errors. This could explain why mortality for this subgroup is very low when reported from experienced centers in contrast to the more equivalent outcomes of Penn class Aa and non-Aa patients in other studies. Efforts should be made to analyze and explain early Penn class Aa mortality to identify modifiable factors amenable to improvement by structured programs. While not surprising, our study findings should be interpreted cautiously, acknowledging the small study sample, retrospective study design, omitted potentially influential variables, and