Near-infrared spectroscopy for neuromonitoring of unilateral cerebral perfusion

Abstract

OBJECTIVES

There is neither consensus regarding which methods of neuromonitoring are adequate and reliable for assessing cerebral cross-perfusion during unilateral cerebral perfusion (UCP) nor are any threshold values defined. The aim of the study was to evaluate the usefulness of near-infrared spectroscopy (NIRS) for the neuromonitoring of right-sided UCP, which is increasingly used for cerebral protection as a consequence of the recent rise in supra-aortic cannulation methods.

METHODS

For the purpose of the study, 122 patients (mean age 67 ± 12 years) who underwent open aortic arch surgery between August 2007 and July 2011 using right-sided UCP with a duration time exceeding 20 min were evaluated. The neuromonitoring consisted of NIRS and pressure measurement in both radial arteries in all patients. Forty-four (36%) patients suffered acute aortic dissection (3 having cerebral malperfusion), and 89 (73%) underwent total or subtotal arch replacement. Logistic regression analysis was used to model neurological adverse outcome (permanent and temporary neurological dysfunctions) as a function of cerebral oxygen saturation and other covariates.

RESULTS

During UCP (mean duration 38 ± 18 min) performed at a constant blood temperature of 28°C, the mean brain oxygen saturation dropped on the non-direct perfused side from 66 to 61% on average, corresponding to 92% of the baseline. In only 1 patient, an insufficient cross-over perfusion was presumed due to an intense drop of the saturation to 15% and was treated by employment of bilateral perfusion. In all remaining patients, the drop was not below 40% and/or 70% of the baseline. In the adjusted analysis, acute aortic dissection could be found as an independent predictor of an adverse neurological outcome (5 permanent, all in acute dissections, and 9 temporary dysfunctions), while there was no association between the occurrence of adverse neurological outcome and the values of regional cerebral oxygen saturation during UCP.

CONCLUSIONS

NIRS seems to be a reliable instrument to recognize a relevant disruption of cerebral cross-perfusion during UCP. A drop of brain oxygen saturation to 40% and/or 70% of the baseline can be considered a threshold value for sufficient cerebral cross-perfusion, at least under the flow and temperature management presented.

INTRODUCTION

Brain protection using unilateral cerebral perfusion (UCP) is gaining increased interest during aortic arch surgery due to its simplicity and clinical safety [1–3]. However, the efficacy of UCP for cerebral protection in complex aortic arch surgeries with prolonged perfusion time is still debated. There is also no agreement about which methods of neuromonitoring are adequate for the assessment of cerebral protection, let alone a definition of reliable threshold values for insufficient cross-over perfusion. In contrast to the limited usefulness of transcranial Doppler, the measurement of regional cerebral saturation with a near-infrared spectroscopy (NIRS) is easy and fast to perform on all patients and can therefore be performed even in emergencies. The aim of the study was to evaluate the usefulness of NIRS for neuromonitoring in patients with prolonged duration of right-sided UCP.

PATIENTS AND METHODS

Between August 2007 and July 2011, a total of 379 patients underwent open aortic arch surgery at our centre. Cannulation of the innominate artery (38) or right carotid artery (314) with right-sided UCP for brain protection was used in 352 patients as is our routine, and, according to the vascular anatomo-pathology, cannulation of the left carotid artery and left-sided UCP was performed in all but one of the remaining patients. The latter patient was the only one in whom bilateral cerebral perfusion (BCP) with cannulation of the supra-aortic arteries on both sides was performed primarily due to complex extra- and intracranial vascular pathologies. For the purpose of the study, only patients with right-sided perfusion were included because this technique is most frequently used for UCP, regardless of whether the innominate, right axillary or right carotid artery is cannulated. In addition, only patients with a UCP duration >20 min were considered in the evaluation because the shorter times might not sufficiently reveal the changes of cerebral oxygen saturation in the non-direct perfused hemisphere. Thus, the study cohort consisted of 122 consecutive patients with right-sided UCP and its duration exceeding 20 min. The patients' mean age was 67 ± 12 years. Forty-four (36%) patients suffered acute aortic dissection, 3 having cerebral malperfusion and 89 (73%) underwent total or subtotal arch replacement. Detailed preoperative patient characteristics are presented in Table 1.

Surgical technique

Depending on the extent of surgery, the thoracic aorta was exposed via a median sternotomy in 117 patients, including a left lateral extension in 3, or clamshell thoracotomy in 5, respectively. The surgical procedures and the operative data are shown in Table 2. The arterial return for cardiopulmonary bypass (CPB) was achieved by isolated cannulation of the right carotid artery or the innominate artery in 107 (87.7%) and 8 (6.6%) patients, respectively. In 7 (5.7%) patients with acute aortic dissection, simultaneous carotid and femoral cannulation with a Y-shaped line was performed to ensure sufficient cerebral and global perfusion [4]. The arterial cannulation technique has been previously described [5]. In short, the innominate artery was isolated within the chest or the right common carotid artery was prepared through a separate approach in the neck along the medial margin of the sternocleidomastoid muscle. In all patients, after heparinization, the exposed segment of the artery was cross-clamped, a longitudinal incision was made and an 8- or 10-mm vascular sealed polyester graft was anastomosed to the artery with a continuous 5.0 polypropylene suture. If additional cannulation of the femoral artery was deemed necessary to ensure sufficient perfusion of the lower body in particular cases with malperfusion, it was done simultaneously in the usual manner, and both arterial lines were connected with a Y-shaped tube for arterial in-flow from one pump [4]. After connecting the arterial line and cannulating the right atrium, CPB was started with a mean flow of 4.6 ± 0.5, with a range 3.0–6.0 l/min (2.2–2.4 l/min/m² of body surface). In all patients, circulatory arrest (CA) under mild-to-moderate hypothermia with UCP was used for arch repair. The technique of UCP has been previously described [1, 6]. In brief, the arch arteries were cross-clamped, and during CA of the lower body UCP was set up at a constant blood temperature of 28°C by simply reducing the arterial flow to a mean flow of 1.5 ± 0.3 l/min.

Cerebral monitoring tools included arterial pressure lines in both radial arteries and NIRS. All measurements were recorded continuously, and for the purpose of the study, the mean values were calculated for each of the five operative phases: 1—before CPB (the NIRS measurement in this phase was a baseline for the relative assessment of tissue oxygen index during all further phases); 2—during CPB with cooling; 3—CA with cerebral perfusion; 4—CPB with re-warming; and 5—after CPB.

Statistical analysis

All perioperative data were prospectively collected. Adverse neurological outcome was defined as a permanent focal neurological deficit confirmed by a neurologist as well as by computed tomography or magnetic resonance imaging, or temporary neurological dysfunction (confusion, delirium, agitation or temporary focal deficits without evidence on computed tomography or magnetic resonance imaging). Univariate associations between preoperative and intraoperative variables and adverse neurological outcome were determined using the t-test, χ² test or Fisher's exact test as appropriate. Multivariate logistic regression analysis was used to determine the independent predictors of an adverse neurological outcome and, especially, to reveal any association between an adverse neurological outcome and the regional cerebral saturation on the left (non-direct perfused) side, and/or the pressure in the left radial artery.

The statistical analysis was performed with the SPSS software version 18.0 (SPSS Inc., Chicago, IL, USA). Categorical variables are reported as frequencies and continuous variables are reported as mean ± standard deviations unless otherwise indicated.

RESULTS

During UCP (mean duration 38 ± 18 min) performed at a constant blood temperature of 28°C, the mean tissue oxygen index dropped on the non-direct perfused side from 66 ± 7.7 to 60.7 ± 8.8%, corresponding to 92.5 ± 13.4% of the baseline (Fig. 1). In only 1 patient, an insufficient cross-over perfusion was presumed due to an intense drop of the saturation to 15% (20% of the baseline) that was combined with very low pressure in both radial arteries (about 18 mmHg on the right and just slightly over 0 mmHg on the left) immediately after commencing UCP. For this reason, bilateral perfusion was employed by additional cannulation of the left carotid artery leading immediately to the recovery of left-sided saturation (Fig. 2). This patient had complex intra- and extracranial pathologies, such as an aberrant right subclavian artery originating from the descending aorta and incompleteness of the circle of Willis (CW) with abnormal anterior and posterior components on the right side. Moreover, in our total experience up to date, in which more than 700 patients were operated on using UCP, this case was the only one with proven insufficiency of cross-over perfusion and has therefore been described in detail elsewhere [7]. Among all remaining patients, the deepest drop of the mean tissue oxygen index was 40% (1 patient) corresponding with 78.4% of the baseline in this case. Altogether, there were 32 patients with a drop of the mean tissue oxygen index to 55% or below. In relation to the value of regional cerebral saturation before CPB, 17 patients revealed a drop of saturation <80% of the baseline with the deepest values between 70 and 71% in 2 patients. The mean pressure in the right and left radial arteries during UCP was 66.6 ± 16.3 and 38.8 ± 10.0 mmHg, respectively.

The neurological condition could be assessed in all but 1 patient (female, 76 years), who died before regaining consciousness. She suffered from a thoracic mega-aortic aneurysm combined with coronary heart disease and died after coronary artery bypass grafting and complete replacement of the entire thoracic aorta through a clam shell thoracotomy. The cause of death was postoperative myocardial infarction due to a heparin-induced thrombocytopenia and coronary graft occlusion. Another 4 patients died during the 30-day postoperative course resulting in an early mortality of 4%. Two of them (both with acute aortic dissection and also preoperative cerebral malperfusion in 1) suffered permanent neurological deficit, which was the primary cause of death.

Altogether there were 5 (4%) permanent neurological deficits, all of which occurred in patients with acute aortic dissection exclusively, and were preceded with preoperative cerebral malperfusion in 3 of them. There were no deviations from a normal pattern of regional cerebral saturation in any patient. The median tissue oxygen index was 59% (range 53–75%) corresponding with 98.1% (range 80.3–123%) of the baseline.

Postoperative temporary neurological dysfunction occurred in 9 patients (4 suffering acute aortic dissection) with a median age of 76 years (range 51–86). The median time of UCP was 29 (range 23–66) min in this group.

Univariate analysis revealed acute aortic dissection, preoperative cerebral malperfusion and male sex to be possible predictors of an adverse neurological outcome, while in the following multivariate logistic regression analysis, acute aortic dissection was identified as an independent predictor of an adverse neurological outcome with an odds ratio of 4.2 (95% confidence interval 1.2–15.0; P = 0.027). In the adjusted analysis, there was no association between the occurrence of an adverse neurological outcome and the values of regional cerebral oxygen saturation, the pressure in the left radial artery, nor the duration of UCP—at least within the time range presented in the examined population.

DISCUSSION

The rationale for the use of UCP is to simplify the perfusion and surgical strategy by avoiding any manipulations on the supra-aortic arteries, with the exception of the artery cannulated. Both the systemic perfusion during CPB and the cerebral perfusion during CA require only one cannula connected to a supra-aortic artery. The switch between systemic and selective brain perfusion can be easily achieved by clamping the supra-aortic arteries proximally to the arterial line and reducing the flow rate. However, the UCP is non-physiological circulation, and the effective collateral network is crucial for sufficient brain protection. The primary collateral pathway is the CW, which provides the most efficient connection between the arterial supplies of both hemispheres. On the basis of anatomical or angiographic studies, it has been revealed that, in more than 50% of the human population, several variations of the CW exist [8–11]. A relevant incompleteness of the CW from a surgical point of view can be expected in about 20% of patients when left-sided UCP with perfusion of the left carotid artery is used and in less than 5% during right-sided UCP with perfusion of the right carotid and vertebral arteries [12]. The clinical and experimental studies reveal, however, that even in such situations, an impaired cerebral flow can be well tolerated, probably by recruitment of secondary extracranial collateral networks [1–3, 12–14]. It is also possible that the thin-caliber vessels of the circle, which appear aplastic or hypolastic on angiographic examination, often widen when the flow is impaired, ensuring sufficient cross-perfusion. Despite the reported clinical safety of UCP in aortic arch surgery, some vascular anatomo-pathologies may lead to failed protection and, consequently, make the assessment of sufficient cross-over perfusion neuromonitoring indispensable [1, 7, 15–17]. This evaluation consists of preoperative diagnostics and intraoperative neurovascular monitoring. The role of preoperative diagnostics is limited because of the reasons mentioned above, and we therefore abandoned preoperative cranial angiography. However, we consider the routine use of neurovascular monitoring to be mandatory [7, 12].

The simplest and also most reliable and practical method is pressure measurement in both radial arteries, revealing the global collateral flow to the left subclavian artery [1, 12]. This can be completed by using other tools, e.g. assessment of regional perfusion with NIRS, the usefulness of which has been evaluated in the series reported here. The method is based on the Beer–Lambert law and the different rates of absorption of near-infrared light by the brain tissue depending on the tissue saturation. The final value of brain tissue oxygen saturation (rSO₂) assessed by NIRS is the arterial and venous mixed saturation, representing the outcome of arterial oxygen delivery and the cerebral metabolic rate by oxygen consumption. NIRS can be installed very fast (only two sensors have to be fixed on the patient's forehead) and is easy to use. However, the procedural differences in the perfusion strategy (flow, pressure and especially temperature) may play a meaningful role in the function of collateral pathways. We are convinced that the moderate hypothermic brain perfusion (not less than 28°C), which is consequently combined with an alpha-stat strategy, is of utmost importance for maintaining cerebral autoregulation and equally distributed perfusion [18, 19]. Furthermore, NIRS is limited to the regional assessment of the frontal lobes and should therefore be completed by additional tools indicating the efficacy of global perfusion. For this, we strongly favour the routine use of blood pressure measurement in both radial arteries. Using this strategy, we did not observe any correlation between adverse neurological outcome and rSO₂, the pressure in the left radial artery, nor the duration of UCP—at least within the time range presented in the examined population. The study reported here not only confirms the efficacy and safety of UCP for brain protection, but also reveals that NIRS seems to be a reliable instrument to assess cerebral cross-perfusion. The first aspect could also be revealed in some studies analysing the use of UCP and bilateral perfusion in similar clinical situations [16, 17], even if there is still a lack of randomized examinations. Hence, it seems that a risk of insufficient cross-perfusion during UCP is very low and that embolic complications (in chronic aneurysms) or malperfusions (in acute dissections) play a more important role in the development of adverse neurological outcomes. The risk of embolic complications can even be slightly increased in bilateral cerebral perfusion because of indispensable manipulations on the supra-aortic arteries. However, even if insufficient cerebral cross-perfusion is very seldom during UCP, the use of NIRS can be very practical. NIRS indicates impaired perfusion by a massive drop of oxygen saturation in such situations, while in the majority of cases, there is only a slight drop of rSO₂ in the contralateral hemisphere confirming safe and sufficient protection. Our experience revealed that, during UCP averaging about 40 min, even a drop of rSO₂ to 40% and/or 70% of the baseline was generally associated with good collateral flow as measured in the left radial artery and therefore was clinically well tolerated.

In conclusion, NIRS seems to be a reliable tool to recognize a relevant disruption of cerebral cross-perfusion during UCP. A drop of brain oxygen saturation to 40% and/or 70% of the baseline can be considered a threshold value for sufficient cerebral cross-perfusion, at least under the perfusion and surgical management presented.

ACKNOWLEDGEMENT

The authors thank Melissa Lindner, Alexandra Metz and Bianca Müller for the assistance in preparing this article.

Conflict of interest: none declared.

REFERENCES