Impact of the Type of Anesthesia on Outcome after Elective Endovascular Aortic Aneurysm Repair: Literature Review

Abstract

The type of anesthesia used during aneurysm repair affects postoperative outcomes for the patient. Although endovascular aneurysm repair (EVAR) appears to improve surgical outcomes, by convention, general anesthesia remains predominantly used. The aim of this study was to compare the impact of the type of anesthesia (ie, locoregional versus general anesthesia) on the outcomes following EVAR. A literature search was carried out using the PubMed search engine to find relevant published articles that compared locoregional and general anesthesia in patients undergoing EVAR. The review of the selected studies showed that although patients in the locoregional group were less medically fit compared with those in the general anesthesia group, there was a reduction in the cardiovascular support required during and after the surgery, postoperative hospital stay, intensive care unit (ICU) stay, and postoperative mortality and morbidity. Although there is no level 1 evidence for or against locoregional anesthesia in EVAR, conventionally, EVAR has been performed under general anesthesia. But this is rooted in tradition rather than evidence. This review suggests that locoregional anesthesia can improve postoperative outcomes following EVAR by reducing hospital stay, ICU stay, mortality, and morbidity, although other factors may also have some influence.

Keywords

anesthesia aneurysm aorta endovascular general local

There is currently no level 1 evidence for or against locoregional anesthesia (LRA) in endovascular aneurysm repair (EVAR). Conventionally, EVAR has been performed under general anesthesia (GA).

In EVAR-2, 47% of the EVARs were performed under LRA,¹ and in DREAM, this figure was 46.1% (39.8% regional and 5.3% local anesthesia).² Within the confines of these large multicenter studies, the 30-day mortality was only 1.7% and 1.2%, respectively, and it seems unlikely that a difference in mortality between LRA and GA will ever be convincingly proven in the face of such low mortality rates. As such, the first portion of this article examines the potential pros and cons of each technique, followed by a review of the existing literature.

Most teams start from the position of using GA as their standard anesthesia and consider LRA only when GA is deemed to be less than optimal. Although the data from EVAR-2 have been criticized, they have raised the question of whether those patients unfit for open abdominal aortic aneurysm (AAA) repair should be treated at all.³ If a patient is only fit for LRA, should he or she have an EVAR at all? There is, however, some evidence that patients whose preoperative morbidity relates to chronic obstructive pulmonary disease rather than cardiac disease may fare somewhat better than expected. A cohort of patients on home oxygen with an AAA > 6 cm treated with open repair or EVAR had a 45% increase in survival at 3 years when compared with a similar untreated cohort.⁴

It is difficult to determine the risk of GA alone apart from during the operation that is taking place. A Japanese study of almost 1.3 million anesthetic episodes concluded that most morbidity and mortality related to the surgery and that only one fatality occurred directly from 100,000 anesthetic episodes.⁵ This anesthetic department study looked at all procedures under GA and was not limited to vascular patients. The physiologic arguments surrounding the benefits, or otherwise, of remaining awake during stressful vascular surgical procedures remain unanswered at present but will undoubtedly be clarified with the publication of the carotid GALA trial. GALA (General Anaesthesia versus Local Anaesthesia for Carotid Endarterectomy) is a multicenter randomized trial that aims to recruit 5,000 patients to assess the impact of the type of anesthesia on postoperative mortality and morbidity.

Indeed, GA may well have a number of theoretical advantages over LRA for EVAR. In patients with short, angulated necks for whom optimum stent placement is imperative, GA will nearly always guarantee a still patient in whom respiration can easily be suspended for critical periods. Many patients find angioplasty balloon inflation an unpleasant experience, and this is replicated when passing a 24F sheath into a small iliac artery. If blood flow beyond the stent deployment device is limited for more protracted periods, for example, during prolonged attempts at contralateral limb cannulation, lower limb ischemic pain can result in a restless patient at just the wrong time. Occasionally, however, a lack of pain and continued movement under LRA can be reassuring, especially when covering the subclavian artery or landing the stent close to the carotid origin. Conversely, an overeffective lower limb epidural may provide unnecessary cause for concern after aortoiliac EVAR or, worse, lead to an assumption that the legs will begin to move once the LRA wears off.

The use of LRA does not remove the need for careful anesthetic monitoring, usually with at least an arterial line and often some central access. Indeed, epidural-induced hypotension may require treatment if there is a concern that the patient may be at risk of spinal cord ischemia. Although blood pressure modulation is possible under LRA, the period of stable hypotension used for optimal thoracic aortic aneurysm (TAA) stent deployment may be less easy to maintain. In the advent of an inadvertent iliac disruption, GA may mask some of the initial early signs but may speed up the process of emergent laparotomy.

In patients with a ruptured AAA, the use of true local anesthesia may have a number of advantages. The induction of GA is often associated with a further decrease in systolic pressure, with potentially serious consequences. The loss of muscle tone can also remove a degree of tamponade, further reducing perfusion pressures. In patients who are only just maintaining an adequate blood pressure or exhibiting signs of instability, initial deployment of a uni-iliac stent can be performed under local anesthesia (LA) and then completed under GA once the patient is more stable. Patients with a ruptured TAA and a large hematoma may become markedly dyspneic once they are laid flat for an LA repair.

Methods

A systematic literature search of the related articles published between January 1994 and December 2008 was carried out by two independent researchers (U.S. and P.D.H.) using the PubMed and Embase databases. The key words used for the literature search included “aneurysms,” “abdominal aortic aneurysm,” “endovascular repair,” and “anaesthesia.” Relevant medical journals were also hand-searched. Each relevant article retrieved had its references searched for missed reports.

The inclusion criteria were published comparative studies and randomized controlled trials in English, in which authors had compared outcomes of LRA and GA in EVAR patients. The minimum number of patients to be reported by a study to be included for analysis was more than 40. The outcome measures that we looked at were the need for postoperative cardiovascular support, 30-day mortality, intensive care unit (ICU) stay, total hospital stay, blood loss, and total operative duration.

Results

No randomized controlled trial was identified, but six comparative studies were identified through PubMed and Embase searches and selected for this review. These studies compared the outcomes of LRA and GA in EVAR patients, in line with the selection criteria. Important results of the included studies are shown in Table 1 and Table 2.

Table 1.

Impact of the Type of Locoregional Anesthesia on Outcome after Elective Endovascular Aortic Aneurysm Repair

Study	Total LRA Patients	Conversion to GA	Postoperative Hospital Stay (d) Mean(range)	Operative Time (min) Mean (range)	Blood Loss (mL) Mean (range)	Fluid Requirements (mL) Mean (range)	Mortality	Morbidity
Henretta et al⁶	47	1	2.1	170 (90–431)	623 (15–2,500)	2,491 (800–6,800)	0	0
Cao et al⁷	64	4	2.6	124	286	0.13 units (blood transfusion)	0	0
Bettex et al⁸	63 (local)	5	3 (3–16) total hospital stay	100			0	0
Bettex et al⁸	8 (epidural)	1	4.5 (4–12) total hospital stay	125			0	0
de Virgilio et al⁹	71			180	2,168			7% (pulmonary)13% (cardiac)
Verhoeven et al¹⁰	170 (local)	2	3.6 ± 3.4 (total hospital stay)	109 ± 30			0	1% (nonfatal cardiac)
Verhoeven et al¹⁰	31 (regional)		5.2 ± 3.3 (total hospital stay)	100 ± 20			0	1% (nonfatal cardiac)
Ruppert et al¹¹	310 (local)	1	3.7 ± 3.1 (1–24)	357.4 ± 436.3 (10–2,000)		357.4 ± 436.3 (10–2,000)	6 (1.9%)	3(cardiac)3 (pulmonary)7 (renal)1 (sepsis)
Ruppert et al¹¹	1,399 (regional)	8	5.1 ± 7.5 (1–106)	643.9 ± 672.1 (100–4,500)		643.9 ± 672.1 (100–4,500)	32 (2.3%)	41 (cardiac)25 (pulmonary)22 (renal)3 (sepsis)

GA = general anesthesia; LRA = locoregional anesthesia.

Table 2.

Impact of the Type of General Anesthesia on Outcome after Elective Endovascular Aortic Aneurysm Repair

Study	General Anesthesia	Postoperative Hospital Stay (d) Mean	Operative Time (min) Mean	Blood Loss (mL) Mean	Fluid Requirements (mL) Mean	Mortality	Morbidity
Cao et al⁷	54	3.2	125	323	0.35 U	0	3%
Bettex et al⁸	20	5.5 (4–16)	172.5		RBCs: 4 (maximum 10 U) (20%)FFP: 4 (maximum 12 U) (20%)Platelets: 2 (12 U) (5%)	0	0
de Virgilio et al⁹	158	1.9 (ICU stay)	210		2,928 mL	Cardiac death: 3%	Pulmonary: 3.8%Major cardiac: 9%Cardiopulmonary: 13%
Verhoeven et al¹⁰	38	5.4 ± 3.3 hospital stay	139 ± 48			1 (3%)	Nonfatal cardiac morbidity, n (%): 1 (3)Respiratory complications, n (%): 5 (13)Renal failure, n (%): 2 (5)
Ruppert et al¹¹	3,838	6.2 hospital stay				4.3% (high-risk GA)

FFP = fresh frozen plasma; GA = general anesthesia; ICU = intensive care unit; RBC = red blood cell.

The first reported series of EVAR under LA was from Illinois in 1999.⁶ The authors used intravenous sedation (a mixture of propofol and midazolam) and LA in a series of 47 consecutive patients suitable for EVAR. The mean American Society of Anesthesiologists (ASA) score was 3.1, with 30% scoring 4. All patients had an EVAR successfully deployed, but one required conversion to GA for a disrupted iliac artery. Although two more patients required retroperitoneal exposure to repair injured external iliac arteries, this was done under local anesthesia. Within 24 hours of graft implantation, 98% of patients were tolerating oral diet and ambulant. The mean time to discharge was 2.1 days, although the authors felt that this would have been shorter if there was no wait for discharge computed tomography. The average operative time was 170 minutes, with a range of 90 to 431 minutes. The estimated blood loss averaged 623 mL (range 150–2,500 mL), and the fluid replacement requirements were 2,491 mL (range 800–6,800 mL). There was no mortality and no 30-day cardiovascular comorbidity; however, three patients (6.4%) had noncardiopulmonary morbidity in the 30-day perioperative period. Although only a quarter of the patients were awake enough to be cooperative during the entire case, most were easily aroused within minutes of stopping the titrated sedative drips. Most of the patients were able to control their own airways, with only three patients requiring an oral airway or other airway protection.

The second major series of elective EVAR in 1999 was from Perugia, which reported outcomes in a nonrandomized series of 61 (54%) patients treated with epidural anesthesia (EA) against 54 (46%) treated with GA.⁷ The choice of anesthesia was left to the anesthetist. EA was performed by placing an epidural catheter at the L2-L3 level. A loading dose of 10 mL of 0.5% bupivacaine or 10 mL of 0.75% ropivacaine, associated with mild sedation with intravenous (IV) midazolam (2 to 4 mg), was used for an epidural block. The intent of the anesthetist was to sedate patients at a Ramsay level 2 (“cooperative, oriented, and tranquil”). Inadequate pain control was treated with further loading until the desired analgesia (at the level of T10) was achieved. Around 75% of each cohort was ASA 3. There was a difference in patient demographics and aneurysm morphologies between the two groups (ie, EA and GA). Similarly, on an intention-to-treat analysis, the authors found no differences in length of total hospital stay, ICU care, fluoroscopy time, or other outcomes between the two groups. Only when the four conversions from the EA group to the GA group were excluded did the need for ICU stay fall significantly (0 vs 5 patients; p = .02) and length of hospital stay (2.5 vs. 3.2 days) become shorter in the EA group. But GA and ASA 4 were found to be independent predictors of prolonged postoperative hospitalization on multivariate logistic regression.

Bettex and colleagues reported a retrospective analysis of 91 consecutive patients in 2001.⁸ Of this group, 63 had true local anesthesia, 8 EA, and 20 GA. Lidocaine 0.5 to 1% to a maximum of 500 mg was used for bilateral inguinal infiltration. The epidural catheter was placed between L3 and L5. A sensory anesthetic level at T10 was established with bupivacaine 0.5%. Intravenous sedation involved a titration of midazolam and/or a continuous infusion of low-dose propofol, supplemented as required with fentanyl, sufentanil, or nicomorphine. The percentage of patients needing ICU care postoperatively was 27% (n = 17), 50% (n = 4), and 70% (n = 14), respectively. In addition, they reported that the LA group needed less inotropic support and that they were given 1,000 mL less intravenous fluid. In addition, the mean hospital stays were 3, 4.5, and 5.5 days, respectively, with a statistically significant difference between the LA group and the GA group (p < .0005). The operating time was significantly less in the LA group compared with the EA and GA groups. Although there was no mention of the AAA complexity in each group, the two groups were comparable with regard to their comorbidities.

In 2002, de Virgilio and colleagues reported on rates of cardiopulmonary morbidity and mortality following EVAR under GA (n = 158) or LA infiltration plus propofol, midazolam, and fentanyl (n = 71).⁹ The type of anesthesia was determined by the surgeon, the anesthetist, or patient preference. The rates of pulmonary complications (GA 3.8% vs LA 7%) or major cardiac problems (GA 9% vs LA 13%) were not statistically significantly different from one another. However, the length of ICU stay was significantly lower in the GA group at 1.9 versus 2.3 days. Conversely, anesthetic time, operative time, and intraoperative fluid infusion were significantly less in the LA group compared with the GA group. When multivariate analysis of the data was performed, the only predictor of complications was the number of preoperative comorbidities rather than anesthetic type. No mention is made of length of hospital stay. Again, it should be noted that this group had an LA procedure in only relative terms.

A joint Dutch-Canadian study from 2005 reported the results of EVAR in 170 LA, 38 GA, and 31 regional anesthetic (RA) cases.¹⁰ Intravenous sedation was used as an adjunct to LA in only 13% of cases, and only two LA cases converted to GA. The technique of LA was very carefully described and had clearly been thought through to minimize patient discomfort from dissection, stent advancement, and limb ischemia. Patients with a body mass index > 30 were excluded from LA. The majority of patients (52%) were ASA 3 or 4. According to AAA classification, the patients undergoing GA had slightly more anatomically difficult aneurysms (p < .01). The number of serious respiratory complications was significantly higher in the GA cohort (13% vs 0 for the LA and RA cohorts). The total operating time was significantly shorter in the LA group than in the GA group (109 vs 139 minutes). The number of days in hospital was significantly fewer in the LA group at 3.6 days compared with 5.4 days for the GA group and 5.2 days for the RA group (p < .001).

An analysis of the EUROSTAR data in 2006 examined the influence of anesthesia type on outcome after EVAR.¹¹ There were 3,838 GA cases, 1,399 (25%) RA cases, and 310 (6%) LA cases. There were significantly more ASA 3 and 4 patients in the RA and LA groups relative to the GA group (67%, 51%, and 45%, respectively). Despite the higher ASA status of the LA and RA groups, there were significantly more systemic complications in the GA group (13%) relative to the RA group (9.5%) and the LA cohort (6.6%). These differences were all statistically significant. The rates of ICU admission postoperatively were 16% for GA, 8% for RA, and 2% for LA; not surprisingly, these differences were statistically significant. The LA cases used the least theater time at 116 minutes, relative to 133 for GA (p < .001). The length of hospital stay was reduced from 6.2 days with GA to 5.1 days with RA and to 3.7 days with LA. It may be tempting to put these differences down to the most difficult cases being performed under GA, but the data do not support this. For GA, RA, and LA, the neck diameters were 23.8, 24.0, and 23.8 mm, respectively. The patients having GA had slightly smaller AAAs at 58 mm rather than 59 mm for RA and LA. The neck lengths were 27, 26, and 28 mm, respectively, with only the 26 mm RA neck being statistically different. Interestingly, the experience of the teams performing was significantly greater in the LA and RA groups, with around 64% of them performing more than 30 EVARs per year as opposed to only 44% in the GA group.

Further analysis of the EUROSTAR registry was done in 2007, which confirmed the above findings.¹² From July 1997 to August 2004, 5,557 patients were enrolled in the registry and divided into two groups (ie, low risk and high risk). ICU stay was significantly less frequent for high LA than high RA and high GA, with high RA having a significant advantage over high GA. Systemic complications were lower for both high LA (9.0%; p = .012) and high RA (10.7%; p < .0001) than for high GA (18.3%). Early death (≤ 30 days) was reduced in high RA (3.0%) versus high GA (4.3%; p = .028); therefore, it is concluded that minimally invasive anesthetic techniques provided significant benefit to high-risk patients.

Conclusions

All of the above studies were nonrandomized trials or extended case series. However, some trends do seem to emerge. Despite the fact that often the LRA groups were less fit than the GA patients, there were consistent reductions in ICU use and length of hospital stay. Postoperative mortality and morbidity were also less in the LRA group. Similarly, there was also a reduction in the cardiovascular support required during and after the surgery. Although shorter hospital stays and ICU stays have also been reported for open repair group by different centers, some of the findings could be due to the skill, interests, and specific techniques of the anesthesia teams rather than simply the differences between GA and LRA. Although EA has potential side effects, such as epidural hematoma, none of the major studies reviewed here reported any side effects. This may have cost-effectiveness implications as both local and epidural anesthesia lead to a reduction in the costs of EVAR performed under LA relative to GA.

Such data are likely to increase the uptake of EVAR. However, most centers start off with EVARs under GA and take up LA once they have sufficient experience with this technique, as is the case in our practice and is our recommendation. This is because of the steep learning curve associated with EVAR. This supports the fact that as larger, more experienced centers perform a greater proportion of their cases under LA, weight is added to calls for centralization of specialist vascular services.

References

10.

11.

12.