Keywords
aorta/aortic - database - kidney
Introduction
Outcome after surgery for type A aortic dissection has been continuously improving
over the last decade.[1] Major conceptual improvements have been the routine use of the right subclavian
artery for arterial inflow, selective antegrade cerebral perfusion, and a trend toward
less extensive core temperatures during hypothermic circulatory arrest.[2] Nevertheless, there is still room for improvement, in particular about neurologic
outcome as well as about any type of preoperative malperfusion syndrome, as this is
known to substantially affect the outcome.[2]
[3]
This study aims to analyze survival, kidney function, and neurologic injury after
surgery for acute type A aortic dissection in a large single-center cohort.
Materials and Methods
Study Cohort and Follow-up Modality
We analyzed 445 patients having undergone surgery for acute type A aortic dissection
in our institution between 2001 and 2013. In 2013 all expected alive patients were
contacted and written informed consent was obtained. In case of inability to reach
an individual, the last documented follow-up available in our records was used. The
local ethics committee approved the study (KEK-ZH-Nr. 2012–0039). The contact algorithm
is shown in [Table 1]. Descriptive characteristics of the patient cohort are given in [Table 2]. Median follow-up was 3.0 (range: 0–12.5) years.
Table 1
Follow-up characteristics of the cohort
Total no. of patients
|
445
|
Sum of follow-up years
|
1,707
|
Follow-up > 3 mo or deceased
|
430
|
Follow-up > 3 mo
|
308
|
Information by patient
|
235
|
Information by general practitioner
|
32
|
Data from CCIF
|
41
|
Deceased patients
|
122
|
No follow-up > 3 mo
|
15
|
Abbreviation: CCIF, computed clinical information system.
Table 2
Descriptive characteristics of the cohort (n = 445)[a]
Demographics
|
|
Age, mean, SD, y
|
62.7 ± 13.3
|
Age > 70 y, n (%)
|
148 (33)
|
Age > 80 y, n (%)
|
35 (8)
|
Female, n (%)
|
122 (27)
|
Chronic health conditions and risk factors
|
Chronic obstructive pulmonary disease, n (%)
|
47/436 (11)
|
Renal impairment (creatinine > 200 µmol/L), n (%)
|
27/439 (6)
|
Diameter ascending aorta > 45 mm, n (%)
|
277/440 (63)
|
Arterial hypertension (RR syst > 140 mm Hg), n (%)
|
296/379 (78)
|
Diabetes mellitus, n (%)
|
21/242 (9)
|
BMI, mean, SD
|
26.8 ± 4.6
|
Preoperative assessment
|
EuroSCORE, additive, mean, SD
|
11 ± 3
|
Redo surgery, n (%)
|
22 (5)
|
Preoperative tamponade, n (%)
|
64/440 (15)
|
Left ventricular dysfunction (LVEF < 55%), n (%)
|
191/420 (46)
|
Left coronary artery affected by dissection, n (%)
|
31/440 (7)
|
Right coronary artery affected by dissection, n (%)
Cardiogenic shock
|
57/440 (13)
45/442 (10)
|
Intraoperative assessment
|
Cross-clamp time (min), median, (IQR)
|
91 (55)
|
Total circulatory arrest time (min), median, (quartile 3)
|
2 (17)
|
Selective cerebral perfusion (min), median, (IQR)
|
28 (20)
|
Antegrade cerebral perfusion, n (%)
|
375/414 (91)
|
Retrograde cerebral perfusion, n (%)
|
12/406 (3)
|
Extent of operation
|
Valve
|
Bentall procedure, n (%)
|
125/445 (28)
|
Resuspension, n (%)
|
138/445 (31)
|
Arch
|
|
Hemiarch replacement, n (%)
|
240/445 (54)
|
Total arch replacement, n (%)
|
50/445 (11)
|
Postoperative assessment
|
Freedom from dialysis, n (%)
|
390/437 (89)
|
Freedom from long-term dialysis, n (%)
|
416/435 (96)
|
Freedom from reoperation, n (%)
|
404/445 (91)
|
Freedom from ischemic stroke, n (%)
|
373/440 (85)
|
Freedom from hemorrhagic stroke, n (%)
|
421/428 (98)
|
Abbreviations: BMI, body mass index; IQR, interquartile range; LVEF, left ventricular
ejection fraction; RR, relative risk; SD, standard deviation; syst, systolic.
a Missing data lead to varying total numbers.
Anesthesiologic Management
Patients did not receive any premedication as for emergency procedure. Anesthesia
induction and maintenance were standardized for all cases. A rapid sequence induction
with midazolam (0.05–1 mg/kg), etomidate (0.2 mg/kg), for hemodynamic unstable patients
or propofol bolus (1–3 mg/kg) for hemodynamic stable patients, with fentanyl (5 µg/kg),
and rocuronium (1 mg/kg) was used. Anesthesia was maintained with Sevorane (1.5–3
vol%) or propofol (2–3 µg/mL), continuous perfusion of remifentanil (0.1–0.5 µg/kg/min),
and additional boli of rocuronium as needed. Anesthetic monitoring included American
Society of Anesthesiologists routine monitors, two to three invasive blood pressures
of the upper and/or lower extremities (via the left and right radial and femoral arteries),
nasopharyngeal, and bladder temperature measurements. A four-lumen central venous
catheter (7 F; Arrow, Reading, Pennsylvania, United States) and sheath (8.5 F; Arrow)
were placed in the right internal jugular vein. At the end of anesthesia induction,
a multiplane transesophageal echocardiography probe was inserted (Philips Medical
Systems, Andover, Massachusetts, United States). Intraoperative, cerebral perfusion
was monitored with continuous bilateral near-infrared oximetry (NIRS-Invos, Covidien,
Mansfield, Massachusetts, United States). The simultaneous performance of a processed
electroencephalogram (BIS, Aspect Medical Systems, Norwood, Massachusetts, United
States) allowed the confirmation of burst suppression before initiation of hypothermic
circulatory arrest. After median sternotomy, heparin (500 IU/kg) was given to achieve
an activated clotting time of more than 350 seconds before cardiopulmonary bypass
(CPB) and 500 seconds during bypass, with additional boli of 5000 to 10,000 IU if
the activated clotting time did not reach the target values. CPB was primed with 10,000
IU of heparin. Simultaneously with heparin, tranexamic acid (30 mg/kg, maximal dose
of 2 g) was administered intravenously. Therefore, besides premedication, anesthesiologic
management was comparable to standardized management as published elsewhere.[4]
Conduction of Extracorporeal Circulation and Myocardial Protection Strategy
Patients were cooled to 20°C nasopharyngeal temperature and between 26°C and 28°C
bladder temperature. Vasodilators such as phentolamine were used to achieve homogeneous
cooling by reducing peripheral vascular resistance. During rewarming, targets were
36°C nasopharyngeal temperature and 35°C core temperature, comparable to strategies
published elsewhere.[4] Cardioplegia was given as to institutional standards. After termination of ECC,
reversal of heparin with a protamine ratio of 1:1 (1 mg protamine per 100 IU heparin)
was performed. An intraoperative autologous transfusion device with reinfusion of
shed mediastinal blood and remaining ECC blood was frequently used. Intraoperative
and postoperative transfusion regimens and coagulation management were guided by in-hospital
standards (institutional algorithm) following rotational thromboelastometry (ROTEM,
Pentapharm GmbH, Munich, Germany) and laboratory analysis of selected parameters of
coagulation.
Definition of Clinical Parameters
Preoperative parameters were defined according to EuroSCORE guidelines.[5] Evaluated outcomes included long-term morbidity and mortality. Morbidity focused
on postoperative renal failure with need for dialysis and occurrence of stroke (ischemic
and hemorrhagic). Preoperative renal impairment was defined as preoperative creatinine
greater than 200 µmol/L. No patient of the entire cohort was on renal replacement
therapy prior to acute type A aortic dissection. Postoperative dialysis was stratified
into need for acute (new onset of dialysis) or chronic renal replacement therapy for
longer than 1 year after surgery or until death. The long-term setting had been appointed
to exclude influences from concomitant effects of the acute setting, for example,
distributive shock, sepsis, and malperfusion. Neurologic injury was defined as any
newly developed sensorimotor deficit persisting at any point of follow-up combined
with a morphologic correlate in cranial computed tomography. This definition of stroke
was formulated to be conservative, considering that not all patients had routine neurologic
consults postdissection. No evaluation of spinal cord ischemia was performed.
Statistical Methods
Freedom from events was analyzed using Kaplan–Meier curves and reported as estimate
at 1, 3, 5, and, if reasonable, 10 years ± standard error. Groups were compared using
the log-rank test. The effect of continuous and binary variables was analyzed using
univariate Cox-regression. Odds ratios (OR) are reported with 95% confidence intervals
(CI). A multivariable logistic model was applied to assess independent preoperative
risk factors for overall outcome. We considered age, renal impairment, and reduced
left ventricular ejection fraction, the preoperative risk factors that were statistically
significant in univariate analysis for inclusion. EuroSCORE levels as a score of different
factors was not considered for multivariate analysis. With regard to fewer events
of postoperative neurologic and renal impairment, no appropriate multivariate analysis
was possible for these morbidity aspects. Analysis was performed using IBM SPSS Statistics
(Version 22. Armonk, New York, United States).
Results
Survival
Overall, 1-, 5-, and 10-year survival rates were 82.8 ± 1.8%, 73.6 ± 2.4%, and 59.3 ± 3.9,
respectively. Preoperative renal impairment (p = 0.011), higher age (p < 0.001), reduced left ventricular ejection fraction (p = 0.001), higher numeric EuroSCORE levels (p = 0.002), longer cross-clamp times (p < 0.001), and longer cerebral perfusion times (p = 0.001) were predictors of mortality at any time point during follow-up ([Table 3]). In the multivariate analysis ([Table 4]) preoperative renal impairment (p = 0.001), higher age (p < 0.001), and reduced left ventricular ejection fraction (p < 0.001) were independent preoperative risk factors for mortality. No significant
distribution difference was observed between patients with or without critical preoperative
status in terms of elevated creatinine levels greater than 200 µmol/L (p = 0.077). Up to 4 years survival curves diverge between patients with and without
preoperative elevated creatinine levels. After 4 years only few patients at risk are
left in the preoperative elevated creatinine group ([Fig. 1]).
Fig. 1 Kaplan–Meier curve of overall survival subject to preoperative renal impairment.
Pts, patients.
Table 3
Univariate analysis of overall mortality
Risk factor
|
OR
|
95% CI
|
p Value
|
Chronic obstructive lung disease
|
1.533
|
0.924–2.541
|
0.098
|
Creatinine > 200 µmol/L
|
2.176
|
1.193–3.97
|
0.011
|
Diameter ascending aorta > 45 mm
|
0.901
|
0.59–1.377
|
0.630
|
Arterial hypertension
|
1.817
|
0.991–3.331
|
0.054
|
Diabetes mellitus
|
1.1
|
0.463–2.608
|
0.83
|
Affected left coronary artery
|
1.369
|
0.751–2.497
|
0.305
|
Affected right coronary artery
|
1.378
|
0.842–2.257
|
0.202
|
Redo surgery
|
1.628
|
0.847–3.129
|
0.144
|
Cross-clamp time
|
1.006
|
1.003–1.009
|
< 0.001
|
Cerebral perfusion time
|
1.015
|
1.006–1.024
|
0.001
|
Impaired left ventricular function
|
0.965
|
0.945–0.986
|
0.001
|
EuroSCORE
|
1.168
|
1.059–1.289
|
0.002
|
BMI
|
1.003
|
0.962–1.046
|
0.88
|
Age
|
1.037
|
1.021–1.054
|
< 0.001
|
Abbreviations: BMI, body mass index; CI, confidence interval; OR, odds ratio.
Table 4
Multivariate analysis of preoperative predictors for overall mortality
Risk factor
|
OR
|
95% CI
|
p Value
|
Creatinine > 200 µmol/L
|
2.858
|
1.532–5.335
|
= 0.001
|
Impaired left ventricular function
|
0.443
|
0.296–0.664
|
< 0.001
|
Age
|
1.050
|
1.032–1.069
|
< 0.001
|
Abbreviations: CI, confidence interval; OR, odds ratio.
The total number of deaths was 120. Detailed in-hospital records from 62 patients
were available. Of these, 41 deaths were aortic and 9 cardiac related. Three were
due to infection and three to bleedings. Six patients had other causes of death. 58
patients died during follow-up in an ambulant setting. No detailed medical reports
or autopsy results were available for these patients.
Dialysis
Overall risk for new dialysis was 6 ± 1.8%, 10 ± 1.7%, and 13 ± 2% within 1, 3, and
5 years after surgery for acute type A aortic dissection. Preoperative renal impairment
(p < 0.001), higher age (p = 0.022), higher numeric EuroSCORE levels (p = 0.022), and prolonged cross-clamp and cerebral perfusion times (p < 0.001; p < 0.001) were predictive for the need of perioperative dialysis ([Table 5]).
Table 5
Univariate analysis of overall new dialysis
Risk factor
|
OR
|
95% CI
|
p Value
|
Chronic obstructive lung disease
|
0.943
|
0.288–3.092
|
0.923
|
Creatinine > 200 µmol/L
|
5.278
|
2.277–12.235
|
< 0.001
|
Diameter ascending aorta > 45 mm
|
0.836
|
0.391–1.785
|
0.644
|
Arterial hypertension
|
0.756
|
0.339–1.687
|
0.495
|
Diabetes mellitus
|
0.704
|
0.199–2.485
|
0.585
|
Affected left coronary artery
|
0.356
|
0.049–2.605
|
0.309
|
Affected right coronary artery
|
0.415
|
0.099–1,732
|
0.228
|
Redo surgery
|
1.735
|
0.530–5.6778
|
0.363
|
Cross-clamp time
|
1.011
|
1.006–1.015
|
< 0.001
|
Cerebral perfusion time
|
1.029
|
1.018–1.040
|
< 0.001
|
Impaired left ventricular function
|
0.996
|
0.962–1.032
|
0.84
|
EuroSCORE
|
1.164
|
1.021–1.326
|
0.022
|
BMI
|
1.023
|
0.986–1.081
|
0.42
|
Age
|
1.028
|
1.004–1.054
|
0.022
|
Abbreviations: BMI, body mass index; CI, confidence interval; OR, odds ratio.
Predictors for long-term dialysis were preoperative renal impairment (p < 0.001) and prolonged cross-clamp and cerebral perfusion time (p < 0.001; p < 0.001), ([Table 6]). No statistical evaluation was possible for patients with chronic obstructive pulmonary
disease (COPD), affected left or right coronary ostium, or redo surgery due to low
event rate. Overall risk for long-term dialysis was 3 ± 0.7% and 5 ± 1.3% within 3
and 5 years, respectively, after surgery for acute type A aortic dissection.
Table 6
Univariate analysis of long-term dialysis (> 1 year after operation or until death)
Risk factor
|
OR
|
95% CI
|
p Value
|
Creatinine > 200 µmol/L
|
15.636
|
3.775–64.765
|
< 0.001
|
Diameter ascending aorta > 45 mm
|
1.053
|
0.193–5.761
|
0.952
|
Arterial hypertension
|
1.534
|
0.189–12.524
|
0.688
|
Diabetes mellitus
|
2.272
|
0.276–18.731
|
0.446
|
Cross-clamp time
|
1.011
|
1.006–1.018
|
< 0.001
|
Cerebral perfusion time
|
1.031
|
1.014–1.048
|
< 0.001
|
Impaired left ventricular function
|
0.987
|
0.936–1.041
|
0.63
|
EuroSCORE
|
1.033
|
0.814–1.311
|
0.79
|
BMI
|
0.993
|
0.888–1.111
|
0.90
|
Age
|
1.006
|
0.971–1.043
|
0.72
|
Abbreviations: BMI, body mass index; CI, confidence interval; OR, odds ratio.
Neurologic Impairment
Overall estimated 1-, 3-, and 5-year freedom from neurologic impairment was 90 ± 1.5%,
87 ± 1.7%, and 84 ± 2.l%, respectively. Higher age (p = 0.045) and prolonged cross-clamp and cerebral perfusion time (p = 0.038; p = 0.007) were predictors for disabling stroke ([Table 7]). Seven hemorrhagic strokes occurred during the complete postoperative period (10%
of all events).
Table 7
Univariate analysis of neurologic impairment
Risk factor
|
OR
|
95% CI
|
p Value
|
Chronic obstructive lung disease
|
1.318
|
0.557–3.120
|
0.530
|
Creatinine > 200 µmol/L
|
0.923
|
0.222–3.826
|
0.912
|
Diameter ascending aorta > 45 mm
|
0.899
|
0.459–1.758
|
0.755
|
Arterial hypertension
|
1.018
|
0.485–2.138
|
0.962
|
Diabetes mellitus
|
0.639
|
0.184–2.219
|
0.480
|
Affected left coronary artery
|
0.236
|
0.032–1.720
|
0.154
|
Redo surgery
|
0.704
|
0.667–2.959
|
0.632
|
Cross-clamp time
|
1.005
|
1.000–1.010
|
0.038
|
Cerebral perfusion time
|
1.016
|
1.005–1.028
|
0.007
|
Impaired left ventricular function
|
1.007
|
0.975–1.040
|
0.67
|
EuroSCORE
|
0.980
|
0.863–1.113
|
0.75
|
BMI
|
0.937
|
0.877–1.001
|
0.061
|
Age
|
1.020
|
1.000–1.040
|
0.045
|
Abbreviations: BMI, body mass index; CI, confidence interval; OR, odds ratio.
Discussion
In addition to classic risk factors, survival after type A aortic dissection is affected
by preexisting renal impairment. Risk factors for any neurologic injury also during
follow-up were age and prolonged cross-clamp and cerebral perfusion time. Preexisting
renal impairment is predictive of any postoperative renal failure.
Overall 1-, 3-, 5-, and 10-year survival rates were estimated 83%, 78%, 74%, and 59%,
respectively. The initial attrition rate is primarily due to the event of acute aortic
dissection per se and related problems in untreated segments of the aorta as well
as to cardiac disease, as indicated from 62 detailed in-hospital medical records from
death patients in our cohort. Most series addressing surgery for acute type A aortic
dissection focus on perioperative outcome and often lack reporting of survival greater
than 4 years.[3]
[6]
[7]
[8]
[9]
[10] The few contemporary trials with longer follow-up show comparable outcomes.[11] However, patient number reported is substantially smaller in most studies.
Renal impairment, higher age, and reduced left ventricular ejection fraction were
independent preoperative predictors of mortality. Impaired renal function has been
confirmed in patients with stroke,[12] coronary artery disease,[13] diabetes,[14] hypertension[15] and in the general population[16] as an independent predictor of mortality. However, no other trial had focused on
preoperative renal impairment in patients with acute type A aortic dissection to date.
We identified preoperative renal impairment as an independent predictor for mortality
in surgically treated patients with acute type A aortic dissection. This finding was
independent from critical preoperative state, so we presume not only the acute impairment
in the setting of acute cardiogenic shock or as an entity of malperfusion syndrome
with potential reversal after correcting the underlying pathological mechanisms but
also the long-term manifestation of renal disease affects survival. We suggest that
the pathophysiologic processes underlying vascular endothelial dysfunction due to
renal disease may explain our findings. This is supported by the aspect that over
the first years survival curves with subject to preoperative renal impairment diverge.
Therefore these findings are different to data reported by Tsai et al. who detected
that acute kidney injury 24h after surgery is an important predictor of mortality.[17] One trial reported preoperative serum creatinine values[18] but found no significant change in mortality with higher serum levels. Most important,
different definitions of renal impairment may have led to these results. Our trial
addressed patients with a KDIGO 2012 class G3b-4 (moderate to severe impaired renal
function), whereas Pagni et al[18] used a creatinine cutoff of larger than 1.4 mg/dL, reflecting a mild to moderate
impaired renal function (KDIGO class G2 to G3a). Lower total numbers (132 vs. 445
patients in our trial) and inclusion of nonsurgically managed patient might be additional
confounding factors.
Our finding that impaired left ventricular function is an independent risk factor
is supported by current literature.[18]
The discussion regarding age and outcome in acute type A aortic dissection has several
aspects. As most studies do not report their exclusion criteria for subjecting or
nonsubjecting elderly patients for surgery, outcomes might be somewhat biased as merely
the biologically fittest will gain access to treatment. This might account for the
fact that advanced age is a risk factor in several reports as it is not in others.
Finally it has to be mentioned that the overall number of octogenarians in all series
is low. Longer cross-clamp time and prolonged cerebral perfusion time reflecting the
extent of disease are prescribed as risk factors previously and consistently in current
literature.[18]
Overall risk for new dialysis was 6% in the first year after surgery and 13% in the
5 years after surgery for acute type A aortic dissection. Risk factors for postoperative
dialysis were preoperative renal impairment, higher age, higher numeric EuroSCORE
levels, and prolonged cross-clamp and cerebral perfusion times. Tsai reported that
11.2% of his patients with surgically treated type A or B aortic dissection received
temporary dialysis during the initial hospital stay after surgery.[19] In a recent report, hypertension, lower limb malperfusion, and sepsis were independent
risk factors for postoperative renal failure,[17] but again only in the acute setting after aortic dissection. Several risk factors
in the initial postoperative period (e.g., sepsis, malperfusion) lead to temporary
need for dialysis, but with sufficient treatment, long-term need for renal replacement
therapy should be avoidable. Of note no trial reporting on long-term dependency of
dialysis after surgery for acute type A aortic dissection is available in current
literature. To differentiate between temporary and prolonged need for dialysis, a
second analysis was done only for patients with prolonged (> 1 year after surgery
or until death) need for dialysis. After 5 years the risk of long-term dialysis was
5%, and therefore substantially lower than the overall risk of new dialysis. Predictors
for long-term dialysis were preoperative renal impairment and prolonged cross-clamp
and cerebral perfusion time. Therefore, beside the extent of the disease, reflected
by cross-clamp and cerebral perfusion time, preoperative renal impairment was responsible
for a key factor of long term morbidity. Beside this, renal failure is an important
risk factor for a broad range of other diseases, so treatment strategies in patients
with postoperative renal impairment are crucial. Therefore blood pressure control
and restoring renal blood flow in acute and subacute setting as well as treatment
of an eventually underlying renal disease are of importance to decrease need for dialysis.
Overall estimated 1- and 5-year freedom from stroke-related neurologic impairment
was 90% and 84%, respectively. The IRAD study group reported a total of 6% of patients
with acute type A aortic dissection presenting with stroke initially.[10] Aim of the surgery for acute type A aortic dissection with cerebral malperfusion
is to restore cerebral perfusion that may resolve neurologic deficits. Pagni reported
an incidence of 13.6% of postoperative stroke, with no definition for stroke given
in the paper.[18] Taken the emergency situation prior to operation into account with patients often
arriving intubated in our tertiary care center and even lacking data about a complete
preoperative neurological status, we focused on postoperative neurologic outcome.
Therefore both preoperative neurologic complications and surgical risk factors have
to be considered. Higher age and prolonged cross-clamp and cerebral perfusion time
were predictors for disabling stroke in our analysis. The prospective analysis of
neurologic dysfunction by Conzelmann and colleagues in GERAADA[20] showed a similar pattern for duration of cardiopulmonary bypass and circulatory
arrest. Other risk factors for postoperative stroke like malperfusion and dissection
of the supra-aortic vessels[20] were not analyzed in our study. Beside age, no other examined preoperative condition
met significance in our study.
Clinical Implications
Trials for treatment of renal failure should include acute aortic dissection as a
composite outcome intended to determine the overall impact of the treatment to investigate
strategies (blood pressure control, improvement of lower limb perfusion) to positively
influence pre- and postoperative renal impairment and lower mortality rate. Especially
in high-risk patients aggressive prevention is substantial and therefore therapy addressing
the most relevant risk factors is necessary. Further demographic changes will lead
to increased event rates during the next years.
Study Limitations
Like any retrospective review, this study has limitations. It is a cohort of unselected
patients and the denominator of patients that declined or were not offered surgery
is unknown. Our technique has evolved over the study period (i.e., cannulation site,
root replacement type, surgical skill, cerebral perfusion) and results might also
be affected by the different surgeons involved. With regard to neurologic impairment,
no evaluation of spinal cord ischemia was performed and no evaluation targeting different
surgical techniques or cerebral malperfusion imaging took place.
Conclusion
In addition to classic risk factors, survival after type A aortic dissection is affected
by preexisting renal impairment. Risk factors for any neurologic injury also during
follow-up were age and prolonged cross clamp- and cerebral perfusion time. Finally,
preexisting renal impairment is predictive of any postoperative renal failure and
therefore important for long term morbidity. Consequently, treatment and prevention
strategies for renal failure during the acute and long-term course after acute type
A aortic dissection are warranted and continuous monitoring of these patients remains
mandatory.