Regional Practice Patterns and Outcomes of Surgery for Acute Type A Aortic Dissection

Abstract

Background

The surgical management of acute Type A aortic dissection is evolving and many aortic centers of excellence are reporting superior outcomes. We hypothesize that similar trends exist in a multi-institutional regional consortium.

Methods

Records for 884 consecutive patients who underwent aortic operations (2003 to 2015) for acute Type A aortic dissection were extracted from a regional Society of Thoracic Surgeons database. Patients were stratified into three equal operative eras. Differences in outcomes and risk factors for morbidity and mortality were determined.

Results

Surgery for Type A aortic dissection is increasing in extent and complexity. Aortic root repair was performed in 16% of early era cases compared to 67% currently (p<0.0001). Similarly, aortic arch repair increased from 27% to 37% cases (p<0.0001). Cerebral perfusion is currently utilized in 85% of circulatory arrest cases, most frequently antegrade (57%). Total circulatory arrest times increased (29 vs 31 vs 36 min, p=0.005), but times without cerebral perfusion were stable (12 vs 6 min, p=0.68). While operative mortality remained stable at 18.9% during the 3 operative eras, there were significant decreases in pneumonia and reoperations (p<0.05). Predictors of operative mortality and major morbidity are age (OR=1.04; p<0.0001), prior stroke (OR=2.09; p=0.03), and elevated creatinine (OR 1.31, p=0.01). Importantly, extent of aortic operation did not increase risk for morbidity or mortality.

Conclusions

Operative morbidity and mortality remains significant for Type A aortic dissection, but lower than historical outcomes. Extent of aortic surgery has increased resulting in adaptive cerebral protection changes in contemporary “real-world” practice.

Acute aortic dissection is a life-threatening emergency that requires expedient surgical treatment (1). Stanford Type A Aortic Dissections (TAAD), involving the proximal thoracic aorta, are associated with a 30% overall in-hospital mortality, with risk of mortality increasing by 1% per hour in the acute phase after emergence of symptoms (2, 3). The in-hospital mortality rate for patients with Type A dissections who do not undergo surgical repair is 59% according to the International Registry of Acute Aortic Dissections (IRAD)(4, 5). However, for patients who have operative repair and are discharged from the hospital, survival rates are as high as 96% at 1-year and 91% at 3-years (1, 3, 5).

Many centers of excellence report improved rates of survival in patients undergoing surgical repair of TAAD over the past two decades (6, 7). These improvements have been due to an evolution in approaches, cannulation strategies, cerebral perfusion, and temperature management (8–10). While endovascular techniques have not yet conquered the ascending aorta, the advancements in arch stabilization strategies and the advent of frozen elephant trunks have resulted in a shifting surgical paradigm for TAAD. Additionally, traditional circulatory arrest at 18 degrees Celsius has been transitioned to partial cooling with circulatory arrest at warmer temperatures with retrograde or antegrade cerebral perfusion to reduce the morbidity and mortality associated with this operation (10–12).

These reports about changing practice patterns and improving outcomes are mostly limited to single center and high volume aortic centers. The objective of this study is to evaluate the evolution of surgical approach and outcomes of acute aortic dissection in “real world” practice utilizing a regional database. We hypothesize that surgical practice has changed to incorporate different repair techniques and cerebral perfusion strategies resulting in a reduction in morbidity and mortality associated with surgical repair of TAAD.

PATIENTS AND METHODS

Patient Data

The Virginia Cardiac Services Quality Initiative is comprised of 18 hospitals and cardiac surgery practices in the Commonwealth of Virginia. The registry captures 99% of adult cardiac surgery cases in the state. VCSQI clinical and costs data methodologies have been described elsewhere (13, 14). The participating hospitals and practices submit data to VCSQI using STS data entry forms utilized for the national database. STS clinical data is paired with hospital patient discharge information.

The de-identified records for 884 patients who underwent aortic operations for acute Type A aortic dissection between 2003 and 2015 were included for analysis. Inclusion criteria were acute aortic dissection with aortic root or ascending aortic replacement, while no exclusion criteria were applied. As patient records were de-identified, the University of Virginia Institutional Review Board exempted this study from review (IRB Protocol #19456).

Measures

Patients were stratified into three operative eras based on tertiles of case volume: Early (n=309, years 2003–2008), Middle (n=316, years 2009–2012), and Current (n=259, years 2013–2015). Patient characteristics, operative trends and outcomes were assessed across the operative eras. Additionally, independent risk factors for mortality and major morbidity were identified. Variable definitions are available for each data version (2.52 – 2.81) in the STS Adult Cardiac Database Data Specifications (15). Consistent with these definitions, operative mortality was evaluated as death within 30 days or in-hospital mortality. Major morbidity included reoperation, prolonged ventilation, permanent stroke, renal failure, and deep sternal wound infection.

Statistical Analysis

Univariate analysis was used to evaluate baseline characteristics, operative trends and short-term outcomes. The Mann-Whitney U Test was utilized for continuous variables and Chi-Square for categorical variables. Hierarchic logistic regression was used to analyze the association between patient characteristics and operative morbidity and mortality. Clustering at the hospital level was accounted for as a random effect while changing practice patterns were evaluated by surgical era as a fixed effect. Variables with less than 10% missing data were included for analysis with missing values imputed by the median for continuous variables and mode for categorical variables. All analyses were performed with SAS Version 9.4 (SAS Institute, Cary, NC) and significance was determined by an alpha < 0.05.

RESULTS

Patient Characteristics

Patient demographics and preoperative characteristics for the overall cohort and by surgical era are presented in Table 1. The median age for patients were 59 years with hypertension being the most common comorbidity present in 81% of patients. Notably 11% of patients had prior sternotomies, which was stable over time (p=0.29). Patients were largely similar across the surgical eras. There were significantly lower rates of dialysis dependent renal failure over time (9% vs 1.6% vs 0.4%, p<0.0001) and increased rates of prior stroke in the current era (5% vs 5% vs 11%). Finally, there were increasing rates of moderate to severe aortic insufficiency (AI; 19% vs 21% vs 38%, p<0.0001). There were no differences in other echocardiographic findings or prior cardiac interventions (Table 1). While only available for the current era, the majority of patients (54%) were transferred from outside facilities. Every VCSQI participating hospital performed Type A dissection repairs with a median yearly volume of 1.8 cases/year, an interquartile range of 1.1 to 7.75 cases/year, and minimum of 0.75 cases/year, maximum of 13.3 cases/year.

Table 1.

Baseline Characteristics

Patient characteristics	Surgical Era			p value
	Early (n = 309)	Middle (n = 316)	Current (n = 259)
Age (years; median, IQR)	59 (48–71)	57 (49–71)	59 (50–68)	0.90
Female	109 (35.3%)	95 (30.1%)	97 (37.5%)	0.15
Hypertension	250 (80.9%)	258 (81.7%)	203 (80.6%)	0.94
Diabetes	29 (3.4%)	46 (14.6%)	34 (13.6%)	0.12
ESRD	8 (9.0%)	5 (1.6%)	1 (0.4%)	<0.0001
Peripheral arterial disease	39 (12.6%)	47 (14.9%)	46 (18.9%)	0.13
Cerebrovascular disease	31 (10.0%)	33 (10.4%)	36 (15.1%)	0.13
Prior cerebrovascular accident	16 (5.3%)	15 (4.9%)	26 (10.9%)	0.01
Chronic lung disease, moderate/severe	23 (7.4%)	24 (7.6%)	11 (5.2%)	0.51
Coronary artery disease	30 (9.7%)	26 (8.2%)	24 (9.3%)	0.80
Prior myocardial infarction	7 (9.7%)	29 (9.2%)	20 (8.1%)	0.87
Heart failure within 2 weeks	25 (8.1%)	26 (8.2%)	26 (10.7%)	0.51
LVEF (%; median, IQR)	55 (45–60)	55 (50–60)	57 (53–60)	0.20
Aortic insufficiency, moderate/severe	56 (18.5%)	58 (20.7%)	85 (37.6%)	<0.0001
Mitral insufficiency, moderate/severe	11 (3.6%)	12 (4.6%)	4 (2.1%)	0.36
Prior cardiac surgery	37 (12.2%)	38 (12.1%)	22 (8.5%)	0.29
Transfer patient	†	44 (60.3%)	139 (53.7%)	0.32

^†Not captured in STS database during this era

CABG = Coronary artery bypass graft; ESRD = End Stage Renal Disease on Hemodialysis

Operative Trends

Patients underwent increasingly complex operations over time as indicated by a number of significant differences in operative variables (Table 2). This included increasing rates of aortic root surgery (16% vs 39% vs 67%, p<0.0001) along with aortic arch repair (27% vs 26% vs 37%, p-0.007). While not captured until STS data version 2.73 (the middle surgical era), 84 (78%) arch repairs were hemi-arch and 24 (22%) were total arch repairs. Consequently, cardiopulmonary bypass (CPB) times increased significantly from a median of 171 minutes in the early era to 176 minutes in the middle and finally 186 minutes in the current era (p=0.003). Cross-clamp times similarly increased (97 vs 106 vs 110 minutes, p=0.04). While the number of patients undergoing circulatory arrest did not significantly differ (80% vs 80% vs 83%, p=0.68), the total time of circulatory arrest increased significantly (29 vs 31 vs 36 min, p=0.005). Cerebral perfusion times are also not available for the early era, but circulatory arrest times without cerebral perfusion nonsignificantly decreased between the middle and current eras (12 vs 6 min, p=0.68). Inversely, cerebral perfusion times trended towards a significant increase (24 vs 29 min, p=0.11). There was a trend towards the current era having higher rates of cerebral perfusion use at 85% compared to 74% in the middle era (p=0.06). Similarly only available in later data versions, the lowest body temperature remained at 20 degrees Celsius for the latter two eras (p=0.84).

Table 2.

Operative Characteristics and Strategies

Operative characteristics	Surgical Era			p value
	Early (n = 309)	Middle (n = 316)	Current (n = 259)
CPB time (min; median, IQR)	171 (133–218)	176 (133–229)	186 (153–242)	0.003
Cross clamp time (min; median, IQR)	97 (73–134.5)	106 (76–157)	110 (80–152)	0.04
Circulatory arrest	57 (79.2%)	253 (80.1%)	214 (82.6%)	0.68
Circulatory arrest (min; median, IQR)	29 (24–42)	31 (21–44)	36 (28–48)	0.005
Circulatory arrest without cerebral perfusion (min; median, IQR)	†	12 (0–27)	6 (0–24)	0.680
Cerebral perfusion used during circulatory arrest	†	46 (74.2%)	180 (84.5%)	0.06
Cerebral perfusion type				0.43
Antegrade	†	21 (45.7%)	100 (56.2%)
Retrograde	†	21 (45.7%)	67 (37.6%)
Both antegrade and retrograde	†	4 (8.7%)	12 (6.3%)
Cerebral perfusion time (min; median, IQR)	†	24 (19–34)	29 (20–39)	0.11
Lowest temperature (Celsius; median, IQR)	†	20 (18–26)	20 (18–24)	0.84
Aortic root repair	49 (15.9%)	123 (38.9%)	174 (67.2%)	<0.0001
Aortic arch repair	83 (26.9%)	82 (26.0%)	96 (37.1%)	0.007

^†Not captured in STS database during this era

CPB = Cardiopulmonary bypass; IQR = Interquartile range; TEVAR = Thoracic Endovascular Aortic Repair

Outcomes

The overall operative mortality rate was 18.9% and did not significantly vary over time (p=0.80; table 3). While rates of major morbidity did not statistically decrease over time, the current era had the lowest rate of major morbidity (59% vs 55% vs 53%, p=0.39). In addition, a number of individual morbidities significantly decreased over time. This included pneumonia (18% vs 9% vs 8%, p=0.0003), reoperation for any reason (23% vs 21% vs 15%, p=0.049) and reoperation for bleeding (11% vs 6% vs 4%, p=0.01). Prolonged ventilation trended towards a decrease from 54% to 52% and 44% (p=0.06). The only complication to increase over time was atrial fibrillation (19% vs 27% vs 32%, p=0.001).

Table 3.

Operative Outcomes

Short-term outcomes	Surgical Era			p value
	Early (n = 309)	Middle (n = 316)	Current (n = 259)
Operative mortality	60 (19.4%)	56 (17.7%)	51 (19.7%)	0.80
Major morbidity‡	181 (58.6%)	175 (55.4%)	137 (52.9%)	0.39
Permanent stroke	34 (11.0%)	30 (9.5%)	26 (10.0%)	0.82
Cardiac arrest	18 (5.8%)	18 (5.7%)	19 (7.3%)	0.68
Pneumonia	54 (17.5%)	29 (9.2%)	20 (7.7%)	0.0003
Prolonged ventilation	167 (54.1%)	164 (51.9%)	115 (44.4%)	0.06
Renal failure	64 (20.7%)	54 (17.1%)	44 (17.0%)	0.40
Renal failure requiring dialysis	38 (12.3%)	35 (11.1%)	36 (13.9%)	0.59
Atrial fibrillation	59 (19.1%)	85 (26.9%)	84 (32.4%)	0.001
Deep sternal wound infection	2 (0.7%)	3 (1.0%)	0 (0%)	0.31
Transfusion	291 (94.2%)	293 (92.7%)	248 (95.8%)	0.31
Reoperation for bleeding	33 (10.7%)	20 (6.3%)	11 (4.3%)	0.01
Readmission	31 (12.4%)	35 (13.3%)	28 (14.3%)	0.84
Discharge to facility	73 (29.8%)	96 (36.4%)	92 (43.8%)	0.01
Postoperative stay (days; median, IQR)	9 (6–17)	10 (6–17)	10 (6–15)	0.74
ICU stay (hrs; median, IQR)	296 (97–434)	268 (132–472)	192 (129–545)	0.90

^‡Major morbidity includes: permanent stroke, cardiac arrest, renal failure, deep sternal wound infection, prolonged ventilation, and reoperation

IQR = interquartile range; ICU = intensive care unit

Risk factors for operative morbidity and mortality

Patient characteristics that were independently associated with operative mortality by hierarchic logistic regression included age (OR 1.04, p<0.0001), prior stroke (OR 2.09, p=0.03), preoperative creatinine (OR 1.31, p=0.01), and CPB time (OR 1.01, p=0.0002; Table 4). Separately, factors associated with major morbidity included age (OR 1.02, p=0.002), hypertension (OR 2.27, p=<0.0001), prior stroke (OR 2.21, p=0.02), and preoperative creatinine (OR 1.59, p=0.002). Surgical era was not independently associated with operative mortality (p>0.05) but there was a trend towards increased risk of major morbidity associated with early compared to current era operations (OR 1.52, p=0.051). A complete list of variables is included in supplemental tables 1 and 2.

Table 4.

Risk factors for mortality or major morbidity

	Operative Mortality			Major Morbidity
Risk Factors	OR	Confidence Limits	p-value	OR	Confidence Limits	p-value
Age (per year)	1.04	(1.02–1.05)	<0.0001	1.02	(1.01–1.03)	0.002
Hypertension	0.96	(0.57–1.63)	0.88	2.27	(1.55–3.33)	<0.0001
Prior cerebrovascular accident	2.09	(1.09–3.99)	0.03	2.21	(1.15–4.25)	0.02
Creatinine (per 1 mg/dl)	1.31	(1.06–1.63)	0.01	1.59	(1.19–2.14)	0.002
CPB time (per min)	1.01	(1.00–1.01)	0.0002	1.00	(1.00–1.01)	0.10
Aortic arch replacement	1.12	(0.73–1.74)	0.61	0.79	(0.60–1.11)	0.18
Aortic root replacement	0.89	(0.55–1.46)	0.65	1.19	(0.82–1.72)	0.37
Era (early vs current)	1.11	(0.64–1.90)	0.71	1.52	(1.00–2.31)	0.051
Era (middle vs current)	0.98	(0.59–1.64)	0.95	1.24	(0.84–1.83)	0.28

CPB = cardiopulmonary bypass

COMMENT

The objective of this study was to identify trends in a regional database that is representative of “real world” practice for surgical repair of TAAD. The number of patients included (884) represents one of the largest analyses of its kind. We found significant variation in practice patterns over time as repair techniques increased in complexity, including higher rates of root and arch repair, and shifting cerebral perfusion strategies that are adapting to the more complex surgeries. Operative mortality remains significant at 18.9%, which is only slightly improved compared to historically reported outcomes. Over the time period in this study, operative mortality has remained stable, but rates of morbidities have improved. Additionally, the extent of aortic surgery was not a predictor of risk-adjusted morbidity or mortality.

Historical operative outcomes for TAAD have demonstrated operative mortality rates of 25–31% (7, 8). The current literature is replete with contemporary outcomes from high volume hospitals or consortiums of aortic centers reporting improving outcomes with mortalities of 5–19% (3–5, 7, 9, 16, 17). These analyses suffer from the simple fact that their aortic teams, high volume aortic surgeons, and surgical and intensive care infrastructure may not be representative of the resources available to a majority of patients suffering TAAD. It is clear that at least some of the impact on improving survival is related to these factors (18, 19). The overall operative mortality rate of 18.9% in this cohort, is consistent with the current literature (6–9, 16–18, 20–22). Mody et al. demonstrated significant improvement in 30-day mortality for Medicare patients from 30.7% in 2000 to 21.4% in 2011, similar to an IRAD analysis where in-hospital mortality decreased from 25.0% in an era starting in 1996 to 18.4% currently (7, 8). While our analysis demonstrated no change in operative mortality, this is likely due to the early era mortality rate of 19.4% being lower than both the IRAD and Medicare studies that demonstrated improvement.

While operative mortality has remained stable at 19% throughout the time period in this study, rates of many complications are declining. This is most apparent in this analysis for reoperation (all cause or bleeding related), pneumonia, and prolonged ventilation. Evaluation of complications after TAAD repair has been less well reported, with no improvement over time demonstrated (20). To our knowledge this is the first study demonstrating improvement in multiple complications in a large cohort, and a trend of improvement in risk-adjusted major morbidity. The rate of major morbidity after TAAD repair is high, over 50% in this cohort. While the nature of aortic dissection inherently makes complications likely, future quality improvement efforts should continue to focus on translating the extensive efforts already underway for cardiac surgery to TAAD repair patients.

The operative approach to TAAD is critical for addressing the intimal tear and stabilizing the root and arch. This study demonstrates that to achieve this goal, the complexity of operations is increasing over time with increasing rates of root and arch interventions. The overall rates of these procedures (39% aortic root and 30% arch repair) are similar to IRAD reported rates of aortic root (32%) and arch (35%) replacement (21). The increasing extent of surgery is supported by evidence that residual dissection will lead to late aneurysmal degeneration (11, 23–25). The increasing rates of arch replacement mirrors current research demonstrating improved neurologic outcomes, although the exact method remains controversial (11). To achieve this goal, the vast majority (78%) of arch interventions were hemi-arch repairs. Meanwhile increasing rates of root operations is supported by evidence that addressing the aortic root during the initial operation can reduce later associated complications and need for reoperation, particularly when involving two more sinuses of Valsalva (23, 26). Increasing the extent of repair to address more complex dissections has been advocated for over a decade, and the results are apparent in this multi-institutional analysis (24).

More extensive aortic repair may improve long-term outcomes in many patients at the expense of longer and more complex operations. However, there is some evidence that this may not translate into increased operative risk (16, 22, 24, 26–28). After risk-adjustment neither root nor arch operations independently increased risk of major morbidity or mortality in this cohort. This risk analysis identified age, prior stroke and preoperative creatinine as the only factors that consistently and independently were associated with both major morbidity and mortality. Both age and renal function were similarly predictive in the IRAD population (29, 30). Incidence of prior stroke suggests cerebrovascular disease, which increases risk for future complications, particularly with circulatory arrest (31). As expected with increasingly complex surgeries, cardiopulmonary bypass time increased throughout the study period, and is weakly associated with mortality (OR 1.01), and this slight association does not extend to major morbidity.

As expected, the total circulatory arrest times increased throughout the study period to accommodate the increasing complexity of surgeries performed. No other trends were statistically significant, but all showed changes that support improving cerebral protection. This includes increasing utilization of cerebral perfusion from 74% in the middle era to 85% currently (p=0.06), as well as longer cerebral perfusion times (24 min vs 29 min, p=0.11). Inversely, circulatory arrest times without cerebral perfusion went from 12 min to 6 min (p=0.68). Additionally, the lowest intraoperative temperature remained at 20 degrees Celsius indicating a continued preference for deep hypothermic arrest. The optimal cerebral perfusion strategy in the setting of TAAD remains controversial, with only clear consensus that prolonged circulatory arrest times without cerebral perfusion is detrimental (32, 33). While this cohort demonstrates improved cerebral protection with a likely increasing reliance on cerebral perfusion and continued deep hypothermia, it is representative of the differing preferences at large with no clear trends for antegrade, retrograde or combined cerebral perfusion approaches.

Limitations of this analysis includes susceptibility to biases associated with all retrospective reviews including selection bias. Additionally, databases contain known coding errors, although the STS adult cardiac database is the gold standard for clinical databases and reports excellent data integrity (34). Finally, the analysis was limited by variables added in later STS data versions, most related to cerebral perfusion and circulatory arrest. Similarly, useful information such as dissection anatomic detail and detailed neurologic outcomes are currently lacking.

In conclusion, type A aortic dissection represents a difficult emergent problem that requires tailoring a surgical approach to the patient and their specific disease. It is becoming increasingly apparent that unaddressed dissection in the root and arch leads to future complications and surgeons are responding accordingly by increasing the extent of their repairs. Furthermore, in this statewide cohort root and arch interventions were not associated with an increase in risk-adjusted morbidity or mortality. The mortality rate remains high at 19% although certain morbidities are improving. Finally, surgeons are adapting cerebral protection practices to account for longer circulatory arrest times with a continued preference for deep hypothermia. Over the past 13 years, surgical practice has evolved to more aggressively address the complex problem of TAAD, which unfortunately continues to have a high mortality rate.

ABBREVIATIONS

CABG

Coronary Artery Bypass Grafting

CPB

Cardiopulmonary Bypass

ESRD

End Stage Renal Disease on Hemodialysis

ICU

Intensive Care Unit

IQR

Interquartile Range

IRAD

International Registry of Acute Aortic Dissections

STS

Society of Thoracic Surgeons

TAAD

Type A Aortic Dissection

TEVAR

Thoracic Endovascular Aortic Repair

VCSQI

Virginia Cardiac Services Quality Initiative