Goal-directed cerebral perfusion in aortic arch surgery: scientific leap or hype?

Abstract

Although significant advancements in cerebral protection strategies in aortic surgery have been achieved in recent years, controversy remains on what constitutes the optimal strategy. Deep hypothermic circulatory arrest alone is a viable approach in many instances, but the need for a prolonged duration of circulatory arrest and increasing case complexity have led to the utilization of adjunctive cerebral perfusion strategies. In this review, we discuss the efficacy of deep hypothermic circulatory arrest and its limitations, the role of retrograde cerebral perfusion and unilateral and bilateral antegrade cerebral perfusion, and the trend towards goal-directed perfusion strategies, all emphasizing the pressing need for randomized clinical trials to better define the optimal strategy.

Introduction

Even in the current era, aortic arch surgery remains highly complex and associated with a non-negligible risk of neurologic dysfunction. Permanent stroke is reported to occur in up to 12% and significant cerebral dysfunction in up to 25% of all-comers.¹ Postoperative permanent neurological dysfunction (PND) is defined as the presence of new focal (stroke) or global (coma) cerebral dysfunction confirmed on neuroimaging, and temporary neurologic dysfunction (TND) is the presence of reversible confusion, agitation, delirium, or a motor deficit with negative radiographic evidence of cerebral injury.² The main considerations in cerebral protection are the use of hypothermia and an adjunctive method of cerebral perfusion to reduce the brain’s metabolic demand and safely extend the ischemic period. Although cerebral protection strategies have undergone significant evolution in recent years, the optimal strategy has not been not defined and an evidenced-based approach remains elusive.

Deep hypothermic circulatory arrest

Since the initial description of the use of deep hypothermic circulatory arrest (DHCA) at < 18°C for total arch replacement by Griepp and colleagues³ in 1975, it has been shown that the safe period of DHCA without additional protective adjuncts is 30 min. Fischer and colleagues⁴ demonstrated that after about 30 min of hypothermic circulatory arrest at 15°C, the regional oxygen saturations (rSO₂) fall below 60%.⁴ Beyond 50 min, cellular anoxia and injury occur with significantly increased rates of neurologic dysfunction.^5,6

In experienced centers, however, some groups have reported excellent outcomes with straight DHCA for periods of up to 40 min. In their series of all-comers undergoing straight DHCA, the Yale group reported excellent long-term results with a mean duration of DHCA of 29.7 ± 8.5 min, with 85.5% of patients undergoing < 40 min and 3.1% > 50 min.⁷ Patients were cooled to a target bladder temperature of 20° C for hemiarch or 18°C for total arch replacement. The stroke rate was 2% in the entire cohort, and significantly higher in patients with DHCA times > 50 min (10.5%, p = 0.05). The rate of TND was 5.1%. The authors noted that patients undergoing elective first-time surgery in non-dissection cases had similar survival compared to those of an age- and sex-matched reference population, and DHCA duration was not significantly associated with poor survival on multivariable analysis. However, as nearly 90% of patients underwent DHCA for < 40 min, the comparatively small numbers of patients undergoing DHCA for > 50 min precludes generalizing these findings to prolonged durations of hypothermic circulatory arres, at which point, adjunctive cerebral perfusion strategies are likely necessary. In a recent analysis of the Society of Thoracic Surgeons (STS) Adult Cardiac Surgery Database of > 12,000 aortic arch repairs utilizing hypothermic circulatory arrest between 2011 and 2014, significant variability in temperature and perfusion strategies were found, with the most common being isolated DHCA (25%), DHCA + retrograde cerebral perfusion (RCP; 16%), and DHCA + antegrade cerebral perfusion (ACP; 14%).⁸ Although this report included both high-volume and less experienced centers, it supported the growing consensus that a lack of an adjunctive protective strategy for cerebral perfusion may result in a higher risk of mortality or adverse neurologic complications compared to hypothermic circulatory arrest alone (odds ratio [OR] = 1.6, p < 0.01), and this has been emphasized in other studies as well.

Adjunctive cerebral perfusion strategies

In a retrospective cohort analysis of the STS database on the outcomes of 7830 adult patients undergoing elective aortic arch or hemiarch replacement between 2014 and 2016,⁹ the authors found that the protective effect of an adjunctive cerebral perfusion strategy became significant, starting in the subgroup of patients undergoing circulatory arrest for > 20 min, with an increasing protective effect with longer arrest times. There was an increased risk of death or stroke associated with increased circulatory arrest times, and this was most pronounced in those receiving > 30 min of straight hypothermic circulatory arrest without cerebral perfusion, while for patients receiving either ACP or RCP, arrest time did not become a significant predictor of death or stoke until after 50 min. Compared to DHCA without cerebral perfusion, the use of DHCA with ACP (OR = 0.65, 95% confidence interval [CI]: 0.52–0.81) or RCP (OR = 0.57, 95% CI: 0. 45–0.71) and moderate hypothermic circulatory arrest with ACP (OR = 0.61, 95% CI: 0.46–0.79) were associated with significant reductions in death and stroke.

Although metabolic activity decreases to 50% of baseline at moderate hypothermia (28°C), significant residual metabolic activity may persist at temperatures as low as 18° C, emphasizing the need for adjuncts to cerebral perfusion to prevent ischemic injury, especially at higher target cooling temperatures.¹⁰ Effective cerebral perfusion provides oxygen and substrates to the brain while protecting against the build-up of toxic metabolic waste during the ischemic period. Implementation of RCP and ACP was developed to provide continuous brain perfusion in conjunction with hypothermic circulatory arrest, however, the available data for both of these techniques have been generally limited to single-institutional analyses, meta-analyses, and national registry data.

Retrograde cerebral perfusion

First described by Ueda and colleagues¹¹ in 1990, RCP theoretically extends the circulatory arrest time by supplying oxygenated blood to cerebral territories.¹¹ The proposed strengths of RCP are its ability to maintain cold cerebral temperatures and its efficacy in washing out particulate and gaseous emboli from the cerebral vasculature, which is a valve-less system. Moreover, compared to ACP, it is considered to be more straightforward to establish by providing an uncluttered field while avoiding potential injury from arterial cannulation. In an institutional analysis of the use of RCP for prolonged periods of DHCP, Lau and colleagues¹² conducted a propensity-matched analysis of patients receiving DHCA for < 50 min versus ≥50 min (n = 48 pairs). No difference in major postoperative complications including death, TND, and PND were found. In the entire cohort, PND occurred in 13 (1.2%) patients and TND in 33 (3.2%). In the unmatched population, there was no difference in the rates of PND between the ≥50 min and < 50 min groups (2% vs. 1.2%, p = 0.623) although the rate of TND was significantly higher in the ≥50 min group (8% vs. 2.9%, p = 0.045). After matching, the rates of operative death (2.1% vs. 0.0%, p = 0.315) and postoperative complications including the incidence of TND (6.3% vs. 2.1%, p = 0.307) and PND (2.1% vs. 0.0%, p = 0.315) were similar in the ≥50 min and < 50 min groups, respectively. This analysis supports the findings of others, demonstrating the safety and efficacy of RCP when used as an adjunct to DHCA for complex aortic arch reconstructions.

Questions remain, however, regarding the neuroprotective effect of RCP. Porcine experiments have indicated that < 0.1% of retrograde perfusate from the superior vena cava perfuses the cerebral capillary beds.¹⁰ Furthermore, the efficacy of RCP may also be impaired by venous valves in the internal jugular veins, which necessitate higher perfusion pressures, and has been demonstrated to result in increased intracranial pressures and poorer neurological outcomes.¹³ The growing trend among aortic surgeons is towards the use of ACP and moderate (20.1°C–28°C) or mild (28.1°C–34°C) hypothermia (MHCA) during circulatory arrest.

Antegrade cerebral perfusion

ACP is thought to provide a more physiologic delivery of blood flow, which facilitates cerebral metabolism and preservation of cell architecture, thereby providing more homogenous cerebral perfusion compared to the smaller regions perfused by RCP.¹⁴ The shift towards warmer temperatures of hypothermic circulatory arrest is accentuated by the desire to reduce the time spent cooling and rewarming on cardiopulmonary bypass (CPB), thus minimizing the risk of hypothermia-induced multiorgan dysfunction, coagulopathy, and transfusion requirements.¹⁵

The feasibility of ACP with moderate to mild temperatures has been well established in multiple reports in the literature.^16–20 In a recent analysis with long-term follow-up, Ahmad and colleagues²¹ demonstrated the versatility and feasibility of using ACP in combination with moderate-to-mild systemic hypothermic circulatory arrest (28.7°C ± 0.6°C) among 587 all-comers presenting for aortic arch surgery over a 15-year period.²¹ ACP was delivered at 30°C using a pressure-controlled protocol with flows of 1.2 ± 0.3 L·min⁻¹. The low overall 30-day mortality, PND, and TND rates of 6%, 6%, and 5%, respectively, for elective and emergency cases encompassing a range of complexities highlight the efficacy of selective ACP in combination with moderate-to-mild systemic hypothermia. Our institutional analysis also supports the efficacy of ACP with MHCA in acute type A dissection patients.¹⁷ Leshnower and colleagues¹⁷ compared 82 patients undergoing type A dissection repair with DHCA to 206 patients undergoing repair with MHCA + ACP. The postoperative stroke rate was not different between the two groups (8.5% in DHCA vs. 8.3% in MHCA + ACP, p = 0.8).

The optimal strategy for ACP delivery, however, has not been established. Generally, unilateral perfusion via the axillary, innominate, or right common carotid artery is initiated at a rate of 8–12 mL⁻¹.kg.min⁻¹ at a mean temperature of 27°C ± 4°C. Bilateral ACP can be initiated via direct cannulation of the left common carotid artery following initiation of circulatory arrest for arch work. Unilateral ACP appears to provide sufficient safety for the majority of patients without significant pathologies of the supraaortic and cerebral arteries. In an institutional analysis of 1000 patients with non-acute aortic dissections (i.e., aneurysm, porcelain aorta, chronic dissection, infection) presenting between 2004 and 2017, with a mean circulatory arrest duration of 23.3 ± 17.2 min under MHCA, early 30-day and hospital mortality was excellent at 1.3% and 2.1%, respectively, and rates of PND and TND of 1.0% and 4.9%, respectively.²² Similarly, data from our own institution analyzing over 400 aortic procedures involving the aortic arch (344 hemiarch and 68 total arch replacements) utilizing unilateral selective cerebral perfusion with a perfusate temperature of 16°C and an average distal arrest time of 30 ± 15 min at 26°C, showed an overall mortality of 7%, and acceptable PND and TND rates of 3.6% and 5.1%, respectively.²³ The efficacy of unilateral ACP and MHCA has also been demonstrated in acute aortic dissection. Tong and colleagues²⁴ identified 203 patients presenting with acute type A aortic dissection undergoing total aortic arch replacement at their institution between 2006 and 2014. ACP was used in all patients, with 82 undergoing unilateral ACP and 121 undergoing bilateral ACP. In the bilateral ACP group, cerebral perfusion was performed via the right axillary artery and left common carotid artery, with flow rate of 10–15 mL⁻¹.kg.min⁻¹ and a perfusion pressure of 40–50 mmHg. In the unilateral ACP group, right axillary artery cannulation was performed for CPB and delivery of ACP. Circulatory arrest time was 24 ± 8 min and similar between groups. Overall 30-day mortality (11.6% for bilateral ACP vs. 20.7% for unilateral ACP, p = 0.075), and prevalence of TND (9.2% vs. 4.7%, p = 0.236) and PND (16.9% vs. 8.4%, p = 0.091) were similar in the unilateral and bilateral ACP groups, respectively. It should be noted, however, that these were not randomized groups, and the authors do not comment on when or why one approach was used over the other.

In general, bilateral ACP is advantageous for patients with carotid artery stenosis, previous stroke, or cerebrovascular anomalies such as an incomplete circle of Willis. Although variations are common, perfusion deficiencies based on anomalous anatomy seldom occur clinically, especially for short circulatory arrest durations.²⁵ In the setting of MHCA, bilateral ACP is thought to provide improved bi-hemispherical cerebral perfusion, but there may be an added risk of arch vessel manipulation over unilateral ACP. A contemporary meta-analysis by Angeloni and colleagues²⁶ of over 5400 patients suggested equal overall mortality and PND rates for unilateral and bilateral selective cerebral perfusion. The observed TND rates, however, appeared slightly higher in the bilateral selective cerebral perfusion group, possibly resulting from additional manipulation of diseased vessels in bilateral cannulation,²⁶ underscoring that bilateral cannulation is to be executed with caution in patients with atherosclerotic disease.

In the current era of ACP, there is no consensus on whether unilateral or bilateral ACP offers superior cerebral perfusion, with the decision remaining according to the surgeon’s preference. In a meta-analysis of 6788 patients receiving either unilateral or bilateral ACP, there was no significant difference in mortality or neurological morbidity. However, the authors found that longer circulatory arrest times were associated with increased mortality in the unilateral ACP group for unclear reasons.²⁷ A review of 17 articles (n = 3500 patients) reported similar neurological injury rates of < 5% for both approaches; however, bilateral ACP was recommended when the expected ACP duration was > 40–50 min.²⁸ As most published series comparing unilateral vs. bilateral ACP have utilized unilateral perfusion only if backflow from the contralateral carotid artery proved to provide collateralization, or if near-infrared spectroscopy monitoring excluded contralateral malperfusion, it is difficult to assess any true differences in the risk profile between the two, based on current literature. Available studies do support unilateral ACP in most cases, except when an extended circulatory arrest time is anticipated.

Historically, cerebral protection during ACP has been based on flow and pressure managements as well as cerebral rSO₂. However, increasingly, the approach has been towards targeting specific thresholds for cerebral oxygen delivery. The University of Colorado group recently published their clinical protocol for goal-directed cerebral perfusion.²⁹ This strategy involves using bispectral index pads to the patient’s forehead to serve as an adjunct monitor of brain activity during cooling, as well as cerebral oximetry pads to measure rSO₂. Sequential pH stat (temperature corrected to arterial blood temperature) and alpha stat (non-temperature corrected) acid-base management techniques are utilized. Selective ACP is delivered at an initial rate of 10 mL⁻¹ηkgηmin⁻¹ and titrated to maintain rSO₂ > 60% and a > 20% reduction from baseline. If increasing the perfusate flow and pCO₂ do not correct any identified rSO₂ asymmetries, a left common carotid cannula is added. While data are limited to institutional analyses, they support the finding that cooling to a lower temperature may not have added benefits, and may add unnecessarily to increased CPB duration and the complications of deep hypothermia.

Retrograde versus antegrade cerebral perfusion

As previously discussed, no consensus has been reached on the role of RCP versus ACP in the literature, and both have demonstrated similar safety profiles. The most recent meta-analyses support these findings.^30,31 Hameed and colleagues³⁰ examined a total of 68 studies (n = 26,968 patients) using network meta-analyses to study outcomes between more than two treatment arms, comparing the effect of DHCA, ACP, and RCP on postoperative stroke and operative mortality after aortic arch surgery. Compared to DHCA, both ACP and RCP were associated with significantly a lower postoperative stroke rate (OR = 0.62, 95% CI: 0.51–0.75 for ACP, and OR = 0.66, 95% CI: 0.54–0.82 for RCP) and operative mortality (OR = 0.63, 95% CI: 0.51–0.76 for ACP, and OR = 0.57, 95% CI: 0.45–0.71 for RCP). Among patients receiving ACP vs. RCP, however, the authors found no differences in rates of PND, TND, or operative mortality. Regression analysis demonstrated that circulatory arrest duration correlated with the neuroprotective effect of ACP and RCP compared with DHCA, so that with an arrest time of > 25 min, the benefits of ACP and RCP over DHCA became significant.

However, some groups have found differences between the two approaches in terms of cerebral protection outcomes. In their analysis of 259 consecutive patients undergoing ascending aorta or hemiarch repair between 2006 and 2014, Perreas and colleagues³² found that relative to RCP, ACP may provide some benefits in rates of TND (risk ratio ACP to RCP: 0.235, 95% CI: 0.079–0.699) and a trend towards decreased 30-day mortality (risk ratio ACP to RCP: 0.333, 95% CI: 0.09–1.23).³² On the other hand, in an analysis of factors associated with acute stroke after type A dissection repair in the STS National Adult Cardiac Surgery Database from 2014 to 2017 (n = 7353 patients), RCP was associated with a reduced risk of stroke compared to no cerebral perfusion (OR = 0.75, p = 0.008) or ACP (OR = 0.75, p = 0.007), and this held true regardless of cannulation approach, nadir temperature, or cerebral perfusion time.³³

These conflicting findings may reflect the lack of standardization in clinical practice. A major limitation towards establishing an optimal cerebral protection strategy for aortic surgery is the lack of randomized clinical trials. Moreover, most studies have relied on only clinical neurological endpoints rather than neuropsychological testing or radiographic imaging analysis, which may be more sensitive indicators of neurologic injury.³⁴ In a recently published pilot analysis in our institution, Leshnower and colleagues³⁵ conducted a prospective randomized controlled trial comparing outcomes between a low-risk patient cohort undergoing elective ascending aortic aneurysm repair using DHCA + RCP (n = 11, mean temperature: 19.9°C ± 0.1°C) versus MHCA + ACP (n = 9, mean temperature: 26.3°C ± 1.8°C).³⁵ The primary endpoint was defined as the composite of stroke, transient ischemic attack, and magnetic resonance imaging evidence of brain injury. Secondary endpoints were TND, neurocognitive deficits as assessed by central nervous system vital signs and serum S-100 levels, a nonspecific biomarker for neuronal injury. RCP was delivered at 1°C at a rate of 300–500 mL⁻¹.min⁻¹ to achieve a pressure of 20–25 mm Hg. In the MHCA + ACP group, ACP was delivered via the right axillary artery with clamp occlusion of the innominate and left common carotid arteries during circulatory arrest. No differences in CPB, cross-clamp, or circulatory arrest times were found between the two groups. All patients received neurologist-adjudicated examinations and brain magnetic resonance imaging prior to discharge. Postoperative National Institute of Health Stroke Scale scores and neurocognitive test results were equivalent between groups. There was one confirmed stroke in each group, and two patients in the MHCA + ACP group experienced TND that had resolved by hospital discharge.

However, diffusion-weighted magnetic resonance imaging demonstrated lesions in 100% of MHCA + ACP patients compared to 45% of DHCA + RCP patients (p < 0.010). Additionally, the number of diffusion-weighted imaging lesions was significantly higher in the MHCA + ACP versus DHCA + RCP group (4 ± 3.5 vs. 1.2 ± 2.1, p < 0.010) with the most common area of lesions in the frontal lobes bilaterally, a pattern of brain injury that is more consistent with embolic phenomena rather than ischemic damage due to hypoperfusion. Although this pilot study is limited by a lack of power and the clinical significance of these silent infarcts remains unclear, these findings challenge the belief that MHCA + ACP and DHCA + RCP produce equivalent neuroprotection in patients undergoing limited arch reconstruction, and support a larger randomized clinical trial to evaluate these two neuroprotective strategies further.

In the absence of robust evidence-based and consensus guideline recommendations, contemporary ascending aorta and arch surgery continues to be characterized by wide variation in cerebral protection strategies and clinical outcomes. To further add to the heterogeneity, most of the single-institutional retrospective studies reported in the literature have been performed in high-volume centers, and it is known that poorer outcomes are associated with low-volume centers that represent the majority of centers in the United States, highlighted by the fact that > 75% of centers reporting to the STS between 2010 and 2014 performed < 6 aortic cases annually. Wide variation remains even in high-volume centers, and it is likely that factors beyond the choice of cerebral perfusion (e.g., surgical expertise, team experience, intensive care unit management) contribute to neurological outcomes,³⁶ which may not be appreciated in the available literature.

Conclusion

Since the initial description of DHCA in the management of aortic arch pathology in 1975, the field of aortic surgery has undergone significant advances. It is increasingly accepted that adjunctive cerebral perfusion strategies including ACP and RCP are necessary, especially with extended circulatory arrest durations. However, there remains immense variability in institutional practice, and controversies persist regarding the optimal strategy for cerebral protection. Importantly, these factors emphasize the growing need for large-scale, interdisciplinary, and multiinstitutional randomized clinical trials, broken down by case complexity, type, and acuity, to address the ongoing areas of contention. It is imperative that such trials incorporate the use of radiographic imaging and neuropsychological testing to diagnose neurologic injury, rather than rely solely on clinical observation. By striving for an evidence-based, goal-directed approach to cerebral protection, we can move away from the hype towards scientific advancements to improve aortic surgery outcomes.