1 Introduction

Arterial carbon dioxide tension (PaCO2) is a potent moderator of cerebrovascular resistance by influence on its vascular smooth muscle tension [1], i.e. hypercapnia induces vasodilation and raises cerebral blood flow (CBF) while the reverse takes place during hypocapnia. CBF and PaCO2 are related linearly between 2 and 10 kPa while the CBF response becomes less steep outside that PaCO2 range [2]. The CBF response to changes in PaCO2, the so-called cerebrovascular CO2 reactivity can be assessed by several methods, e.g. with near infrared spectroscopy (NIRS) determined cerebral oxygenation (ScO2), the brain’s CO2-reactivity is 2.1 μmol L−1 kPa−1 oxygenated hemoglobin [3].

For perioperative evaluation NIRS is user-friendly for indirect assessment of changes in CBF but an influence from superficial tissue oxygenation has to be acknowledged [4]. Also it is important that NIRS detects the lower limit for cerebral autoregulation [5, 6] and thereby allows for interventions aimed to maintain ScO2 and thereby improve postoperative outcome [7–12]. The CO2-reactivity, however, varies among patient groups, e.g. presenting with liver disease [13], atherosclerosis [14], or sepsis [15]. Moreover, with reperfusion of ischemic organs or limbs, hypercapnia is developed to an extent that it can become difficult to control CBF or ScO2 by adjusting ventilation and thereby PaCO2 [13, 16].

For patients undergoing open abdominal aortic aneurysm repair, cross clamping of the aorta below the renal vessels reduces cardiac output (CO) and end-tidal CO2 (EtCO2) decreases if ventilation is not lowered to compensate for attenuated need of CO2 elimination and, accordingly, ScO2 may become affected [17]. Conversely, with reperfusion of the lower body, vasodilatory substances including CO2 are liberated to the circulation and provoke cerebral hyperperfusion as illustrated by a ~50 % increase in CBF [17, 18] that can lead to neurological complication and even death [19]. Following the observations by Liu et al. [17] the department intend to maintain ScO2 throughout open AAA repair by adjusting ventilation upon clamping and reperfusion of the aorta and this retrospective study evaluates, for forty-four patients, how well that ventilatory strategy is followed. Thus, we evaluated to what extent ScO2 was kept within reasonable limits (±15 %) and whether eventual changes in ScO2 during open AAA repair were related to those in EtCO2 and thus ventilation.

2 Methods

Data were obtained from the anesthetic charts for patients undergoing open AAA repair (2010–2013). All patients in that period were considered eligibility with endovascular repair of the aneurysm as the only exclusion criterion. In total forty-four patients were enrolled for review that enables us to detect subtle changes (<1 ± 0.5 %) in ScO2 (Table 1). Evaluation of data was initiated in accordance with guidelines provided by The National Committee on Health Research and following approval by the Local Ethical Committee (H-2-2012-114) and the Danish Data Protection Agency (2009-41-3617; 2012-41-1250). All patients gave informed consent to participate.

Table 1 Characteristics of patients
Full size table

From the anesthetic charts we recorded heart rate (HR) and mean arterial pressure (MAP) obtained from a radial catheter in the arm with the highest non-invasively determined blood pressure before catheterizing the artery. The arterial wave form was exported from the blood pressure box to the Nexfin apparatus in which cardiac output (CO) was calculated by Modelflow® (Nexfin; BMEYE, Amsterdam, The Netherlands). Genders, age, height, weight of the patient were provided to the Nexfin prior to monitoring along with the anatomical site of the catheter. ScO2 (Invos 4100 Cerebral Oximeter, Somanetics, Troy, MI, USA) was recorded along with ventilation and EtCO2 as a substitute for arterial CO2. No threshold for either cerebral deoxygenation or hyperoxygenation exist for patients undergoing AAA repair, but significant cerebral deoxygenation was defined as a >15 % reduction in ScO2 relative to baseline [20] while a >15 % increase in ScO2 was considered to represent cerebral hyperoxygenation [21].

For postoperative pain treatment an epidural catheter (Th 8-9) was placed under cover of local anesthesia and a test dose of lidocaine (1 %; 4 ml) with adrenaline (5 mg ml−1) was used to secure that the catheter was in place. A bolus dose of bupivacaine (3–5 ml, 0.5 %) was administered after induction of anesthesia followed by infusion of bupivacaine (4 ml h−1). Anesthesia was induced by propofol (1–2 mg kg−1) and fentanyl (1 μg kg−1). Mask ventilation was established and cisatracurium (0.1–0.15 mg kg−1) facilitated oral intubation and anesthesia maintained by propofol (0.08 mg kg−1 min−1) and remifentanil (0.3–0.4 μg kg−1 min−1) (n = 32) or by sevoflurane (n = 12). Ephedrine (10 mg), calcium chloride (5 mmol L−1) or phenylephrine (0.1 mg) was administered intravenously to raise blood pressure above 60 mmHg [22]. No vasodilatory agents were administered and in case of hypertension, administration of propofol, remifentanil or sevoflurane could be considered to be increased. In general the choice of a particular anesthetic technique for lowering pressure is normally a decision made by the anesthesiologist but this did not appear to have occurred in the patients included for review. Blood products were administered to maintain hematocrit above 30 % (Table 1).

Variables were noted corresponding to three phases of the operation (dissection, aortic clamping and reperfusion) and changes evaluated by Friedman’s repeated measures analysis of variance on ranks, followed by a Tukey–Kramer post hoc test. If any differences between groups were detected we applied a Wilcoxon signed rank test. Differences between patients with or without inhalation of sevoflurane were evaluated using Kruskal–Wallis one way analysis of variance on ranks, followed by Dunn’s post hoc test. Correlations between variables were evaluated by the Spearman method and a P value <0.05 was considered statistical significant (SAS Institute Inc., Cary, NC, USA; Systat Software GmbH., Erkrath, Germany). Hemodynamic variables are presented as median change relative to baseline (interquartile range) while demographic data and perioperative administrations of vasopressors are presented as median (range).

3 Results

In Table 1 demographics, co-morbidity, duration of the surgery, and the perioperative blood loss are presented. Administration of ephedrine, calcium chloride and phenylephrine is listed in Table 2.

Table 2 Administration of vasopressors
Full size table

By clamping of the aorta ventilation was lowered (−0.8 l min−1; −1.2 to −0.1; P < 0.001) to make EtCO2 and thereby ScO2 stable. HR, MAP and CO did not change significantly (Fig. 1). Peak inspiratory pressure (16 cm H2O; 16–17) and positive end-expiratory pressure (5 cm H2O) were kept stable. Conversely, during reperfusion of the aorta ScO2 increased 2 % (−1 to 4; P < 0.001) as EtCO2 became 0.5 kPa (0.1–1; P < 0.001) elevated although ventilation was increased by 1.8 l min−1 (0.9–2.7; P < 0.001) with an increase peak inspiratory pressure from 16 to 17 cm H2O (P = 0.003). At the same time MAP decreased 8 mmHg (−15 to −1; P < 0.001) although CO increased 0.6 l min−1 (0.3–1.0; P < 0.001) while HR was stable.

Fig. 1
figure 1

Individual changes in near infrared spectroscopy determined cerebral oxygenation (ScO2), cardiac output (CO), minute ventilation, mean arterial pressure (MAP), and end-tidal CO2 (EtCO2), during dissection, aortic clamping, and reperfusion of the aorta during open aortic aneurism repair (n = 44). Solid line indicates median

Full size image

Three patients experienced a >15 % reduction in ScO2 during aortic clamping although ventilation was reduced by 0.5–2.7 l min−1 that reduced EtCO2 by 0.1–0.6 kPa. Conversely, one patient demonstrated a >15 % increase in ScO2 upon reperfusion of the aorta and for that patient ventilation was not increased. Thus, ScO2 was kept within ±15 % relative to baseline in 91 % of the patients (40 out of 44). Changes in ScO2 related to those in EtCO2 (Spearman r = 0.41; P = 0.0001) (Fig. 2).

Fig. 2
figure 2

Near infrared spectroscopy determined cerebral oxygenation (ScO2) and end-tidal CO2 tension (EtCO2) during open abdominal aorta aneurism (AAA) repair (% changes from baseline; ±SEM) (Spearman r = 0.4; P = 0.0001; n = 88). Number of patients indicated

Full size image

There were no differences in hemodynamic variables or ventilatory setting between patients who were exposed to intravenous or inhalation anesthesia. One of three patients who demonstrated a low ScO2 inhaled sevoflurane while the patient who demonstrated an increase in ScO2 was exposed to intravenous anesthesia.

4 Discussion

This retrospective cohort study on how well ScO2 is maintained by adjusting ventilation during open AAA repair finds that ScO2 was kept within an acceptable range in 91 % of the patients by reducing of ventilation by ~1 l min−1 during aortic clamping and increase it again by ~2 l min−1 upon reperfusion of the aorta. In support, ScO2 was related to EtCO2 and in the patient who demonstrated a marked increase in ScO2 upon reperfusion of the lower body, ventilation was not increased. According to the anesthetic charts reviewed the present patients were exposed to less marked changes in cerebral oxygenation than those studied by Liu et al. [17] for whom no ventilatory adjustments were made.

With clamping of the aorta below the renal arteries CO was stable although the lower part of the body is not or only marginally perfused. Thus, the need for CO2 elimination is reduced and that affects CBF if ventilation is not reduced [17, 18]. Liu et al. [17] found that such unintended hyperventilation appears with a 13 % reduction in EtCO2 and a consequent reduction in MCAvmean by 12 % and in ScO2 by 4 % relative to the dissection phase. Although this reduction in ScO2 is small, some patients may demonstrate a significant reduction in ScO2, but for the twelve patients studied by Liu et al. [17] the individual values were not reported. In three of the present patients ScO2 was lowered to what we consider the threshold for cerebral deoxygenation [20] although ventilation was reduced by 0.5–2.7 l min−1. Yet for two of these patients there was a reduction in EtCO2 by ~0.5 kPa and therefore the reduction in ventilation has been inadequate.

Reperfusion of the aorta liberates CO2, lactate and other vasodilatory substances such as prostacyclin to the circulation [23] and EtCO2 increases, e.g. by 44 % after reperfusion of the aorta if ventilation is not adjusted [17]. We report one event of cerebral hyperoxygenation and in that patient no ventilatory adjustments were made. Also inadequate ventilatory adjustment is likely to explain the cerebral hyperperfusion demonstrated by an increase by 45–53 % (in some patients by 104 %) in MCAvmean within 20 min after reperfusion of the aorta [17, 18]. Liu et al. [17] identified a relative increase in ScO2 by 9 % upon reperfusion while the increase was only by 3 % in this study where ventilation was adjusted to aim for at stable EtCO2.

Cerebrovascular disease contributes to death after elective AAA repair (7 %) and is even more significant in patients presenting with a ruptured aneurysm (13 %) [24]. However, for the early postoperative period after AAA repair neurological outcome is not evaluated in regard to changes in CBF and in turn ScO2. In patients undergoing carotid endarterectomy, symptoms of cerebral hyperperfusion including headache, vomiting, confusion, seizures, and hemorrhage are reported with an increase in MCAvmean by 100–300 % or a >15 % increase in ScO2 [21, 25, 26]. A cerebral hyperperfusion syndrome can manifest immediately or up to 30 days after carotid endarterectomy and if left untreated mortality is high [26]. We do not expect patients undergoing AAA repair to develop a cerebral hyperperfusion syndrome as the increase in ScO2 after open AAA repair lasts typically less than half an hour. However, even brief episodes of hyperoxygenation may affect Starling forces within the cerebral capillaries and result in endothelial damage. In general, adequate oxygen delivery to the brain during anesthesia secures rapid recovery as exemplified after thoracic aortic repair [10] as for patients undergoing cardiothoracic [7–10, 27], orthopedic [11], or major abdominal surgery [12]. Accordingly, also patients undergoing AAA surgery may benefit if ScO2 is maintained stable.

The danger of a high cerebral perfusion is illustrated in patients with end-stage liver disease [13, 19]. During reperfusion of the grafted liver patients are subjected to cerebral hyperperfusion that may persist into the postoperative period and lead to brain death [19]. Despite adjustment of ventilation during liver transplantation surgery, it remains difficult to keep ScO2 stable [16] as ScO2 seems to increase more in liver compared AAA repair patients for a given change in EtCO2. Thus, for patients with liver disease other vasodilating substances than CO2 may affect the cerebral vasculature [28, 29]. As determined by 133Xenon clearance CBF demonstrates an increase by 20 % with reperfusion of the liver [13] and the increase in PaCO2 may mitigate the CO2-reactivity because of near-maximal cerebral vasodilatation, i.e. the linear relationship between CBF and PaCO2 becomes less steep with high PaCO2 values [2]. As demonstrated for the AAA patients, control of ventilations can secure EtCO2 and thereby ScO2 because these patients, in contrast to those with liver disease, demonstrates a CO2 reactivity similar to healthy subjects [3] (Fig. 2).

In the reported patients, ephedrine was preferred by the anesthesiologist in the dissection phase whereas calcium chloride was administered typically during aortic clamping (Table 2). CO has been reported to decrease by 2.5 l min−1 with a slight increase in MAP (by 6 mmHg) upon aortic clamping [17], however, in the present study the administration of ephedrine and calcium chloride may have prevented such a reduction in CO [22]. The use of vasopressors was balanced during reperfusion of the aorta. Phenylephrine reduces ScO2 by vasoconstriction within extra-cerebral tissue [4, 30] that may have caused false positive events of cerebral deoxygenation, but none of the three patient with cerebral deoxygenation received phenylephrine. As phenylephrine is administered frequently during reperfusion, we cannot exclude that ScO2, as an expression of CBF, may be an underestimate due to skin vasoconstriction. However, we bear in mind that the impact of phenylephrine on ScO2 seems to depend on preload to the heart [31] and skin blood flow may already be near-maximal comprised by release of catecholamines with reduction of the central blood volume [32].

Limitations of this retrospective observational study are to be mentioned. Hemorrhage amounted to 2.2 l in the presented cohort of patients that along with fluid loss may compromise organ perfusion, however, as blood products were continuously administered to secure hematocrit >30 % we find that unlikely. Twelve patients inhaled sevoflurane, which, in contrast to other inhalation anesthetics, does not have a dose related depressive effect on the cerebral autoregulation [33]. Combination of propofol and remifentanil anesthesia preserves cerebral autoregulation and in the studied population no differences between hemodynamic variables among the two groups were observed. Also hypotension develops in up to 95 % of AAA patients in response to a so-called mesenteric traction syndrome [34] and both manifestations could affect CBF and in turn ScO2. The duration of ScO2 below the considered threshold (>15 %) was not noted. Although no data are available for patients undergoing AAA repair, it seems that the area under the threshold correlates to outcome during surgical aortic arch repair [10]. EtCO2 is a trend monitor for continuously assessment of PaCO2 and EtCO2 is typically 0.5 kPa lower than PaCO2, but we acknowledge that this ratio differs with changes in pulmonary circulation, volume of dead space, and ventilatory strategy applied during anesthesia [35, 36]. Diminished cerebrovascular CO2 reactivity may result in a linear relationship between ScO2 and MAP, however, the individual CO2 reactivity was not assessed before surgical incision. Also this study did not address eventual postoperative cerebral complication in relation to deviations in ScO2 and for that purpose a larger population of patients is in need.

In summary, this retrospective observational study finds that cerebral hyperoxygenation during reperfusion of the aorta during open AAA repair is dominated by an increase in EtCO2 as an indication of PaCO2. With a ~1 l min−1 decrease and a ~2 l min−1 increase in ventilation during clamping and reperfusion of the aorta, respectively, cerebral oxygenation was kept within ±15 % relative to baseline, which is a similar deviation going from, e.g. a supine to standing position [37]. Yet for some patients control of EtCO2 was inaccurate, and we can only speculate whether significant deviations in ScO2 affects outcome after open AAA repair.