| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Eur J Cardiothorac Surg 2006;29:720-728
© 2006 Elsevier Science NL
a Division of Cardiac Surgery, Ospedali Riuniti, Strada di Fiume 447, 34100 Trieste, Italy
b Division of Management Accounting, Ospedali Riuniti, Strada di Fiume 447, 34100 Trieste, Italy
Received 2 December 2005; received in revised form 31 January 2006; accepted 2 February 2006.
* Corresponding author. Address: 22, vicolo Scaglioni, 34141 Trieste, Italy. Tel.: +39 040 3994856; fax: +39 040 3994995. (Email: bartolo.zingone{at}aots.sanita.fvg.it).
 |
Abstract |
|---|
 Top Abstract 1. Introduction2. Materials and methods3. Results4. Discussion5. Limitations6. ConclusionsReferences |
|---|
Key Words: Aorta • Stroke • Ultrasound
|
1. Introduction |
|---|
|
TopAbstract1. Introduction2. Materials and methods3. Results4. Discussion5. Limitations6. ConclusionsReferences |
|---|
Epiaortic scanning was applied selectively, as recommended by Wareing et al. [6], when first introduced in our practice early in 2001. As a natural consequence of our satisfaction with its informative potential and the ease of use it was then extended to all patients with the expectation of further improvements of neurologic outcomes. The aim of this study was therefore to retrospectively assess the impact of epiaortic scanning on perioperative stroke rates following cardiac operations and the relative merits of the regular versus the selective application policy.
|
2. Materials and methods |
|---|
|
TopAbstract1. Introduction2. Materials and methods3. Results4. Discussion5. Limitations6. ConclusionsReferences |
|---|
Patients operated upon in the period January 2000 through August 2004 were retrospectively subdivided into three consecutive groups. Group A included 366 patients operated during the 12 months immediately preceding the introduction of EAS. In group B (n
= 1116) EAS was planned for patients 
70 years old or in any suspicion of aortic pathology as suggested by calcifications on preoperative X-ray imaging or intraoperative appearance of ascending aorta. In group C (n
= 690) EAS was intended for use in all patients.
Deaths occurring at any time during the hospital stay or anywhere within 30 days since surgery were categorized as early deaths. Stroke was defined as a new focal neurologic deficit or coma appearing during the awakening time from anesthesia (early stroke) or at any time later during the postoperative stay (late stroke), and at least partially evident more than 24 h after its onset. Patients were routinely assessed by trained intensivists although, when neurologic injury was suspected, they were further evaluated by a consulting neurologist and eventually submitted to cerebral computerized tomographic (CT) scanning. Other definitions conformed to EuroSCORE [12] criteria.
Epiaortic scanning was performed with a 7.5-MHz probe enveloped in a sterile plastic sleeve and a Sonos 2000 machine (Hewlett Packard Co, Andover, MA, USA) as soon as the pericardium was opened. Images were interpreted by the surgeon and recorded for off-line analysis. Early in the series a qualified echocardiographer assisted in evaluating the images in real-time, and was subsequently available for off-line review of unusual findings. Scan data were coded in the divisional database as: 1 = normal, 2 = uniform thickening 1–3 mm, 3 = plaques 4 mm, and 4 = protruding ulcerated atheromas of any size with or without mobile components. Codes were inputted separately for each of 12 aortic segments obtained by dividing the aorta in proximal, middle, and distal quadrants. The aorta and the scan findings were categorized as abnormal in the presence of 1 segment graded 3.
2.1 Surgery
Cardiopulmonary bypass was performed with open-circuit membrane oxygenation and filtered arterial line and cardiotomy reservoir. The body temperature was allowed to drift except for cases requiring circulatory arrest under deep hypothermia. Cold blood cardioplegia was administered during a single cross-clamp period. Flooding the mediastinum with carbon dioxide was introduced halfway this series for open-heart procedures. Air was purged before closing the cardiotomies and by suctioning the ascending aorta after the end of CPB. TEE was available whenever valve repair was contemplated, in which case it was also used to assist the air purge procedure. Gentle palpation of the ascending aorta was employed with utmost caution in group A patients and abandoned thereafter.
The usual end-hole Argyle aortic cannula (Sherwood Medical, Tullamore, Ireland) was replaced by a blind-ended multiple side-holes Soft-flow cannula (Terumo Cardiovascular Systems, Ann Arbor, MI, USA) for all patients 70 years old since early 2004. As an alternative to the standard cross-clamp, an endo-aortic Foley or Pruitt (LeMaitre Vascular Inc., Burlington, MA, USA) balloon catheter was employed in selected instances with extensive, grade 3–4 ascending aortic changes or a preoperatively diagnosed porcelain aorta. The use of tangential aortic clamping for proximal venous grafts connections has been exceedingly rare in this series (see below), as we have subscribed to a single-clamp policy long before this study was started.
Alterations of surgical techniques, and the reasons for them, were coded out of a set of pre-specified options (Table 2). Coronary operations without CPB (OP-CABG) or on CPB without aortic cross-clamping (BH-CABG) accounted for 103 and 40 cases, respectively, though only 60 and 37, respectively, were performed in that manner due to aortic pathology and therefore counted as ‘alterations’. Extended use of arterial conduits (69% of all grafts were arterial in this series) made the necessity of venous connections to aorta fairly uncommon. Performing the proximal anastomosis under circulatory arrest or by an automated connector, or using a thoracic artery as a source let very few cases require tangential aortic clamping though, in that case, it could be safely established on the basis of previous EAS mapping.
|
2.3 Analysis
Outcomes of interest were separately investigated, both in correlation with EAS policies (groups A, B, and C) and, while entirely disregarding group membership, with EAS having actually been performed or not. Analysis was additionally performed in patients submitted to isolated CABG, to verify whether general results could be reproduced in a more homogeneous model. Stroke rates in particular were first tabulated across scan findings in the whole series. Next, they were analyzed by policy groups A, B, and C on an intention to treat principle, both in the whole series and in the CABG cohort. Limited to the CABG cohort, observed total stroke rates were eventually compared to expected estimates, separately for each policy group first; next, disregarding group membership, according to performing or not EAS; and last, among those CABG cases with scan data available, according to the severity of aortic disease. Ten patients (eight in the isolated CABG cohort, and two, five, and three from groups A, B, and C, respectively) missed aortic pathology code values in spite of scan-guided tactic changes recorded in their operative notes. One of them, from group A, suffered a late stroke. Calculations were performed both by handling them as either scanned or not—with similar outputs (data available on request). Here, they are counted as scanned where the code values have no bearing (Figs. 1 and 2
, Table 2, and where observed stroke rates were compared to expected estimates) while in Table 4 they are categorized as ‘not available’ as they actually are.
|
|
|
|
3. Results |
|---|
|
TopAbstract1. Introduction2. Materials and methods3. Results4. Discussion5. Limitations6. ConclusionsReferences |
|---|
|
70 years old and 80 (12%) of the remaining younger patients) and 616 patients from group C (298 (90%) of those aged 70 years and 318 (89%) of the younger ones). At least one aortic segment was graded 3 in 213 (19.1%) of patients in the selective scan policy group B and 148 (21.4%) of patients in the regular scan policy group C, for an overall rate of 16.6% in the whole series. The prevalence of abnormal scan findings (1 segment graded 3) was less in patients younger than 70 compared to older ones, i.e. 19% versus 42.9% (p
< 0.0001), although by no means negligible. Intraoperative tactic changes were more frequent when EAS was performed on a regular basis compared to either of the two remaining policies (Table 2 ). Table 3 breaks down EAS groups’ outcomes by surgical procedures.
|
3.3 Stroke rates
The incidence of early, late, and total strokes were 0.8% (n
= 18), 0.9% (n
= 19), and 1.7% (n
= 37), respectively. Coma was the initial manifestation in 20 patients and was followed by a focal syndrome in six of them. A cerebral CT scan was available in 12 of the 20 patients with coma and in 14 of the 17 patients with a focal syndrome only, and was compatible with an embolic mechanism except for a single watershed late event.
Stroke contributed importantly to the death rate, as 14 patients with stroke died versus 92 dying among those without stroke (37.8% vs 4.3%, p < 0.0001). The risk of dying was almost evenly distributed among early and late stroke patients (8 of 18, or 44.4% and 6 of 19, or 31.6%), and both event rates were significantly greater than in the absence of stroke (p < 0.0001).
Details of stroke rates in reference to scan findings are shown in Table 4
. Close to half the number of strokes occurred in patients in whom EAS had not been performed. Most of the outcome differences were confined to patients 70 years old (Fig. 1).
An early stroke was observed in six patients with grossly pathologic scans. One of them had a grade 3 atheroma in the posterior wall of the distal aorta, and underwent CABG with no changes performed. The remaining five patients had the cross-clamp moved away from the usual site (n = 3), or the aorta was occluded with an endo-aortic balloon (n = 2). The aortic cannulation and the vein graft anastomosis were moved away from the usual sites once each. Confirmatory computed tomographic findings of an ischemic lesion were obtained in all of them. Of the two patients suffering an early stroke despite normal EAS findings, one had undergone a standard CABG procedure despite having three aortic segments with grade 2 atheromas. The second one was an 82-year-old lady submitted to a seemingly uneventful aortic valve procedure.
Early strokes were significantly more frequent in association with an abnormal aorta (Table 4: 1.7% vs 0.3%, p = 0.02). By multivariable analysis a no-scan policy (OR = 4.0, 95% CL 1.4–11.4) and extracardiac arteriopathy (OR = 3.0, 95% CL 1.1–8.0) were independently associated with early stroke. For patients submitted to EAS only, finding at least one segment graded 3 or greater was the only predictor of early stroke (OR = 5.9, 95% CL 1.2–29.3).
Stroke was noted after normal awakening from anesthesia (1–32 postoperative days, mean 6.3 ± 7.7, median 4) in 19 patients. It occurred more frequently in elderly patients compared to younger ones, and then where EAS was not employed, though all differences could be explained by chance alone (Fig. 2). A likely cause or mechanism could be tentatively identified in 13 cases (Table 5 ).
|
0.03). In the CABG cohort (n = 1463), the stroke rate fell from 3.4% in group A to 0.5% and 1.7% in groups B and C, respectively (Table 3, p = 0.002), though once more the difference was statistically related to the gain obtained in the group B compared to the group A (p = 0.001). Adjustment for McSPI did not alter these findings. The risk-adjusted stroke rates were 2.8%, 0.6%, and 2.0% in the groups A, B, and C, respectively, with the difference (p = 0.01) once again due to the A versus B contrast alone. Both groups B and C, with no difference between them, showed an early stroke rate lower than group A (Table 3: 0.4% and 0.5% vs 1.9%, p = 0.02).
3.5 Stroke rates versus McSPI estimates, CABG cohort
The observed stroke rate was similar to the risk estimated by the McSPI model (3.4% versus 3.9%) in the no-scan policy group. On the other hand, it was significantly lower in the two scan groups pooled together (0.9% vs 2.8%, p
= 0.001), with the selective scan group contributing most to the difference.
The stroke rate was lower than expected in the 745 CABG patients not submitted to scan, though only probably significantly so (1.3% vs 2.6%, p
= 0.09). The difference turned out to be significantly larger in the 718 CABG patients submitted to scan (1.4% vs 3.4%, p
= 0.01). Limiting the analysis to the 710 CABG cases with scan data available the observed stroke rate was 1.3%, irrespective of whether they presented at least one aortic segment graded 3 or not, to be contrasted with the expected 4.5% rate in the presence of abnormal scan findings (n
= 237, p
= 0.03) and the 2.7% rate in their absence (n
= 473, p
= 0.11).
Patients’ subsets, analytical methods, and main findings are summarized in Table 6 .
|
|
4. Discussion |
|---|
|
TopAbstract1. Introduction2. Materials and methods3. Results4. Discussion5. Limitations6. ConclusionsReferences |
|---|
Ultrasonographic imaging has greatly contributed to improve our understanding of the role of ascending aortic and aortic arch atherosclerosis as a cause of acute neurologic injury. Broadly speaking, TEE has unveiled the aortic origin of the most so-called cryptogenetic strokes and, when used intraoperatively, unmasked the role of aortic atheromas as a source of embolic strokes [21]. Surgeons do have, in addition, the unique opportunity for direct ultrasound imaging the ascending aorta, an option exhibiting superior diagnostic accuracy compared to both TEE and aortic palpation [4,5]. Based on these reports and others [6,8,22,23], we felt therefore encouraged to adopt epiaortic scanning in the hope of improving our ability to detect aortic atherosclerosis and prevent aortogenic emboli. The most evident consequence of this practice change was a four- to fivefold increase of the maneuvers apt to avoid manipulating areas of atherosclerotic aorta. Our rates of technical changes, 19% and 26% in the selective and regular scan groups, respectively, are fairly close to the 14% reported by Wareing et al. [6] and the 25.6% by Hangler et al. [8]. Whether these measures can lead to improved neurologic outcomes, however, is a different question.
As a matter of fact, the ability of epiaortic ultrasonography to help decreasing the risk of perioperative stroke has not been clearly demonstrated as yet. In a series of 1200 cardiac procedures including CABG, valve, and aortic surgeries with simultaneous carotid endarterectomy in selected instances, Wareing et al. [6] reported a 1.1% incidence of stroke in patients with no or mild aortic disease though, in cases with moderate or severe aortic atherosclerosis, they observed a stroke rate of 3% and 4.8%, respectively. Their experience was closely replicated by van der Linden et al. [22] in a consecutive series of 921 elective patients submitted to EAS before CABG or valve surgery, with a 1.3% stroke rate rising to 8.7% in patients with atherosclerotic aortic disease. Of note, a number of scan-guided technical changes were reported for a sizeable proportion of patients in both these series.
Clinical results obtained by EAS-guided strategies to date have thus seemingly failed to fully neutralize the risk of perioperative stroke [6,22], and one can merely speculate that the event rate might have been even higher in the absence of such a structured surgical strategy. Actually, Duda et al. [23] did show a favorable impact of EAS driven tactic changes in a small series of patients undergoing isolated CABG procedures, which only points to the importance of defining the surgical model selected to assess the impact of both epiaortic scanning and consequent tactic changes on neurologic outcomes. In fact, previous studies suggest that better outcomes are obtained with off-pump CABG procedures compared to operations performed on cardiopulmonary bypass [8,24], though much of the difference can be explained by the limitation of aortic manipulation [25]. Conversely, avoiding cardiopulmonary bypass and aortic cross-clamping is not an option during valve surgery with or without associated CABG, thus possibly explaining the higher risk of stroke recorded in that setting [16,18].
In an attempt to make our finding more generalizable we selected our population sample by excluding a very small proportion of cases so as to avoid excess confounding. Additionally, we separately analyzed the outcomes in the whole series and in the isolated CABG cohort, also risk-adjusting for death and stroke by two external models [12,13]. We eventually distinguished early and late strokes in order to capture any specific effect of the interventions on outcomes. Notwithstanding the limitations of non-randomization, we believe that our results come somewhat closer to demonstrate that EAS-guided tactic changes can help lowering the stroke rate and, more specifically, can favorably impact upon the risk of intraoperative events due to atheroembolism from aortic atherosclerosis.
In our whole series, the total stroke rate decreased following the introduction of epiaortic ultrasonography, and the same was observed in the isolated CABG cohort. The risk reduction was greatest for early events, and consistently so across the different procedural groups, strengthening the link between intraoperative maneuvers and outcome. The improved outcome associated with EAS in the isolated CABG cohort was confirmed following risk-adjustment by the McSPI model. In addition, using the McSPI estimates as a reference, the observed stroke rate turned out to be lower than expected for patients submitted to scan and, at variance with previous data [6,22], that came irrespective of abnormalities in the ascending aorta. Notably, epiaortic ultrasonography exhibited a lesser impact when used in a regular fashion, quite a contradictory finding at first sight. The paradox was likely the consequence of independent, counter-current variation in the rate of late strokes. In fact, when early strokes alone were considered, the extent their rates decreased was remarkably similar for the two different scan policies, both in the whole series and in the isolated CABG cohort.
Highlighting the impact of epiaortic scanning on early stroke rates stems from the assumption that they might best express the consequences of aortic-generated emboli. Early stroke rates consistently fell, both in the whole series and in the CABG cohort, following the introduction of epiaortic scanning. Admittedly, the risk of early stroke increased in association with abnormal scan findings, although the extent it did so was considerably lower than previously indicated [6,22]. Multivariable analysis of predictors for early events confirmed the beneficial albeit incomplete effect of scan-guided interventions.
The residual risk of early stroke in our patients submitted to epiaortic scanning might have been the result of insufficient alterations of the surgical technique. Wareing et al. [6] have previously suggested that minor alterations alone can be of limited help in preventing perioperative events. Although both the sensitivity of the diagnostic technology and our interpretation of ultrasonographic images cannot be completely ruled out as contributing factors, most early strokes occurred in spite of some alterations of the usual cross-clamping maneuver, or the lack of it. The small number of events precludes further insight, but it is our suspicion that insisting with a traditional approach for fear of the greater risk of more radical maneuvers may be a way of fiddling with disaster. On the other hand, the safety of techniques alternative to aortic cross-clamping, including the replacement of the atherosclerotic aorta, remains arguable [10], and more experience is required to support such a choice in the face of lesser degrees of aortic atherosclerosis.
Some of the strokes occurring after a normal emergence from anesthesia might still be related to intraoperative events. For what it's worth, in the three patients with late strokes and abnormal aortic scan both the elapsed time and the specific circumstances made that relation most unlikely. The same arguments would also hold for most of the patients lacking scan data, limiting the weight and the impact of potentially underestimating aortogenic strokes.
Old age effectively discriminated higher risk patients to be screened by epiaortic scanning, to the point that nothing could apparently be gained by extending that strategy to patients younger than 70. Since surgeons freely performed epiaortic scanning at any suspicion of aortic disease even with the selective scan policy, and because of the occasional unexpected finding of severe disease in younger patients, we would be uneasy in subscribing to any age exclusion criterion at this stage. As a matter of fact, almost half the patients less than 60 years old had at least some minor degree of ascending aortic atherosclerosis in our series, while 19% of those younger than 70 presented with at least one aortic segment graded 3.
|
5. Limitations |
|---|
|
TopAbstract1. Introduction2. Materials and methods3. Results4. Discussion5. Limitations6. ConclusionsReferences |
|---|
At the time of this study, screening for carotid artery disease was based on the presence of carotid bruits. There was no strict protocol for the performance of combined carotid and cardiac procedures, and some elective patients were referred for preliminary carotid surgery as well. We therefore felt quite comfortable in excluding a small number of patients undergoing simultaneous procedures and avoid altogether excess confounding.
Early strokes were used as a proxy for emboli from aortic atherosclerotic disease. This of course introduces both the potential for false attributions and excess estimates on one hand, and for loss of sensitivity on the other hand. Case-wise analysis identified only two patients who might have suffered an early stroke for causes other than aortic disease. Similarly, for the majority of late strokes a plausible explanatory mechanism, different from aortogenic embolism, could be found. Alternatively, using all strokes as an aggregate end-point would require an understanding that it is a homogeneous end-point with a single set of potential predictors and etiologies, which of course is not. Admittedly, there can be no perfect criterion to discriminate aortic emboli from all others and we decided that the selected cut-off would be the best possible approximation.
Neurological assessment of our patients was performed by trained intensivists with the back-up of a qualified neurologist if deemed appropriate. It is therefore possible that minor neurological derangements went underdiagnosed.
Hypoperfusion is a potential mechanism of perioperative neurological injury resulting in ‘watershed’ brain lesions. In this series we found only one patient with such a condition after a late stroke event. It should be noted, however, that three patients with late stroke following cardiopulmonary resuscitation and five patients with early stroke died before a CT scan could be performed potentially leading to underdiagnose hypoperfusion.
|
6. Conclusions |
|---|
|
TopAbstract1. Introduction2. Materials and methods3. Results4. Discussion5. Limitations6. ConclusionsReferences |
|---|
|
References |
|---|
|
TopAbstract1. Introduction2. Materials and methods3. Results4. Discussion5. Limitations6. ConclusionsReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  G. Djaiani, M. Ali, M. A. Borger, A. Woo, J. Carroll, C. Feindel, L. Fedorko, J. Karski, and H. Rakowski Epiaortic Scanning Modifies Planned Intraoperative Surgical Management But Not Cerebral Embolic Load During Coronary Artery Bypass Surgery Anesth. Analg., June 1, 2008; 106(6): 1611 - 1618. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 F. Filsoufi, P. B. Rahmanian, J. G. Castillo, D. Bronster, and D. H. Adams Incidence, Topography, Predictors and Long-Term Survival After Stroke in Patients Undergoing Coronary Artery Bypass Grafting Ann. Thorac. Surg., March 1, 2008; 85(3): 862 - 870. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. C. Anyanwu, F. Filsoufi, S. P. Salzberg, D. J. Bronster, and D. H. Adams Epidemiology of stroke after cardiac surgery in the current era. J. Thorac. Cardiovasc. Surg., November 1, 2007; 134(5): 1121 - 1127. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. Zingone, E. Rauber, G. Gatti, A. Pappalardo, B. Benussi, G. Forti, U. Tognolli, and M. Gabrielli Diagnosis and management of severe atherosclerosis of the ascending aorta and aortic arch during cardiac surgery: focus on aortic replacement Eur. J. Cardiothorac. Surg., June 1, 2007; 31(6): 990 - 997. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| ANN THORAC SURG | ASIAN CARDIOVASC THORAC ANN | EUR J CARDIOTHORAC SURG |
| J THORAC CARDIOVASC SURG | ICVTS | ALL CTSNet JOURNALS |
