Changes in the sublingual microcirculation following aortic surgery under balanced or total intravenous anaesthesia: a prospective observational study

Ischemia/reperfusion (I/R) injury is common in patients undergoing aortic clamping for vascular surgery and leads to systemic inflammation and organ dysfunction [1‐3]. The production of pro-inflammatory molecules and oxidative stress increase especially in the reperfusion phase and are responsible for microvascular alterations similar to those observed during sepsis [4‐6]. These include an impairment in blood flow, capillary shunting with reduction in microcirculatory density and increased perfusion heterogeneity, resulting in a mismatch between oxygen delivery and consumption [4, 5]. The severity of microcirculatory dysfunction increases along with the duration of the ischemic phase [2, 3, 7].

A number of studies explored how drugs can modulate microvascular perfusion in different disease states. In septic patients with impaired blood flow auto-regulation, vasoactive agents may have limited potential for microvascular recruitment [8]: the use of norepinephrine titrated to reach a mean arterial pressure (MAP) > 65 mmHg was unable to improve microcirculatory perfusion during septic shock [9, 10]. Dobutamine increased intestinal microvascular blood flow in animal models of sepsis [11] but failed to restore capillary perfusion in patients with septic shock [12]. Conversely, after cardiac surgery a rise in MAP by norepinephrine infusion induced an increase in splanchnic oxygen extraction without altering the intestinal mucosal perfusion, possibly because of autoregulation phenomenon [13‐15].

Anaesthetics can affect microvascular perfusion as well. Propofol causes vasodilation by stimulating nitric oxide production: this may result in microvascular shunting with reduction in capillary density, increased blood flow heterogeneity and reduced tissue oxygen delivery [16‐18]. Volatile anaesthetics also cause dose-dependent vasodilation [19, 20]. In anesthetised dogs, desflurane, unlike isoflurane and halothane, was able to maintain myocardial, hepatic, intestinal and skeletal muscle blood flow [21]. In patients undergoing thoracoscopic surgery, desflurane-remifentanil anaesthesia was associated with better microvascular perfusion as compared to propofol-remifentanil anaesthesia [22]. In female patients undergoing breast surgery, the use of sevoflurane, unlike propofol, was associated with a significant decrease in the capillary filtration coefficient, suggesting a lower risk of interstitial oedema and tissue perfusion alteration [23]. Microcirculatory dysfunction was found in different surgical settings despite the optimization of systemic hemodynamic parameters [24] and was associated with the development of post-operative complications [25].

We aimed to verify whether desflurane might have a better impact on the microcirculation as compared to propofol in patients at risk for I/R injury due to aortic surgery. The goal of this study was to explore changes in sublingual and peripheral muscle microcirculation in patients undergoing elective open abdominal aortic aneurysm repair under balanced or total intravenous anaesthesia (TIVA).

The study was approved by our local ethic committee of Azienda Ospedaliera Universitaria “Ospedali Riuniti” of Ancona in Italy (ClinicalTrials.​gov identifier: NCT03510793, www.​clinicaltrials.​gov, principal investigator: Prof. Abele Donati, date of registration: April 27th 2018, retrospectively registered). A written informed consent was obtained from all patients before enrolment. This was designed as a prospective observational study on 40 patients undergoing elective open infrarenal abdominal aortic aneurysm repair with prosthetic aorto-aortic or aorto-bisiliac bypass under general anaesthesia. The study was performed at Vascular Surgery of the Azienda Ospedaliera Universitaria “Ospedali Riuniti” of Ancona in Italy between September 2013 and November 2017.

Inclusion criteria were: elective infrarenal abdominal aortic open repair; use of a protocol of intraoperative hemodynamic goal-directed therapy; American Society of Anaesthesiology (ASA) class ≤ III. Exclusion criteria were: age < 18 years, pregnancy, endovascular aneurysm repair, presence of infections, trauma, emergency surgery.

Patients received balanced anaesthesia or TIVA based on the attending physician’s decision, resulting in two study groups of 20 patients each. The choice of the anaesthetic regimen was based exclusively on the physician preference and was taken before and independently of the patient’s enrolment, the anaesthesiologist was unaware of baseline microcirculatory measurements. The patients’ distribution between the two groups was initially due to chance; the enrolment was then stopped in the balanced anaesthesia group once the required sample size (n = 20) had been reached, in order to achieve the same number of patients in the TIVA group. Patients undergoing balanced anaesthesia received desflurane (with an inspiratory oxygen fraction of 0.4–0.5) and remifentanil, administered as a target-controlled infusion (Minto model). TIVA was performed with a target-controlled infusion of propofol (Schnider model) and remifentanil (Minto model). In all patients, anaesthetic depth was monitored with spectral entropy and the anaesthetic dose was titrated in order to maintain a state entropy value of 35 to 45%, resulting in an age-adjusted minimum alveolar concentration of 0.8 for desflurane and an estimated effect-site target concentration (Ce) of 2–4 mcg/ml and 2.5–5 ng/ml for propofol and remifentanil, respectively. Neuromuscular blockade was achieved with a bolus of 0.6 mg/kg rocuronium at anaesthesia induction; additional boluses (10 mg) were provided based on train-of-four neuromuscular monitoring. Haemodynamic parameters were monitored in all patients with Flotrac/Vigileo (Edwards Lifesciences); arterial and central venous blood gases were monitored according to routine clinical practice. A protocol of intraoperative goal-directed therapy was applied in all patients for haemodynamic optimization, according to standard care in our unit [26]. This includes: first, stroke volume optimization by fluid therapy (targeting an increase in stroke volume < 10% after a 200 ml fluid bolus over 5 min); second maintenance of a MAP > 70 mmHg by titrated vasopressor administration; third, maintenance of a cardiac index (CI) ≥2.5 ml/kg per m² by titrated inotropic therapy [26].

Measurements were collected at baseline (before induction of anaesthesia) and at end-surgery (before anaesthesia discontinuation) and included haemodynamic parameters, arterial and central venous blood gases and assessment of the sublingual microcirculation and peripheral muscle tissue oxygenation and microvascular reactivity.

Figure 1 shows the study flow diagram, with details on the number of patients evaluated for eligibility and reasons for exclusion. The two groups did not differ for age, gender distribution, comorbidities, ASA score, and intraoperative data including clamping time, fluid balance at end-surgery and vasoactive agent requirements (Table 1). Changes in haemodynamics and blood gases from baseline to end-surgery are reported in Table 2. MAP and systemic vascular resistance index (SVRI) decreased in both groups. The CI increased in patients undergoing balanced anaesthesia, while it remained substantially stable in the TIVA group. The two groups showed similar changes in haemoglobin (Hb), central venous oxygen saturation (ScvO₂), arterial lactate levels and arterial base excess.

Changes in microcirculatory and NIRS-derived parameters are shown in Table 3. The TVD and PVD for small vessels increased at end-surgery in the balanced anaesthesia group, while the small vessel MFI and PPV remained unaltered in both groups (Fig. 2). At both time points, the TVD and PVD for medium vessels were lower in the TIVA group as compared to those in the balanced anaesthesia group (Table 3). Changes in microvascular density were negatively correlated with changes in SVRI (TVD for small vessels: r = − 0.479 p = 0.002; PVD for small vessels: r = − 0.451 p = 0.003). These correlations remained if considering the two groups separately, although appearing weaker in the TIVA group (TVD for small vessels: r = − 0.453 p = 0.045, PVD for small vessels: r = − 0.362 p = 0.116) as compared to those observed the balanced anaesthesia group (TVD for small vessels: r = − 0.526 p = 0.017, PVD for small vessels: r = − 0.515 p = 0.020). No correlation was found between microvascular parameters and CI, MAP or mean norepinephrine dose during surgery. In all patients, a negative correlation was found between changes in SVRI and mean norepinephrine dose (r = − 0.387, p = 0.022).

The StO₂ did not vary from baseline to end-surgery in either group (Fig. 3). The StO₂ downslope 1 increased in the TIVA group at end-surgery (indicating a slower desaturation rate in the initial part of the ischemic phase), while it remained stable in the balanced anaesthesia group (Fig. 3). The StO₂ downslope 2, delta-downslope and StO₂ upslope did not vary in either group (Table 3). The AUC StO₂ increased at end-surgery in the balanced anaesthesia group (Fig. 3). The THI tended to decrease and was significantly lower at end-surgery in the TIVA group, while it remained stable in the balanced anaesthesia group (Fig. 4).

In this prospective observational study on 40 patients undergoing elective open abdominal aortic aneurysm repair with an intraoperative goal-directed haemodynamic optimization, we aimed to explore whether different anaesthetic regimens could have a varying impact on microvascular perfusion in a condition of potential I/R injury. Our first finding is that both balanced anaesthesia and TIVA were associated with overall maintenance of microvascular perfusion and tissue oxygenation. Patients receiving balanced anaesthesia showed an increase in microvascular density, which was inversely related to a reduction in systemic vascular resistance. In patients undergoing TIVA we observed a minor reduction in the skeletal muscle tissue oxygen extraction rate during a VOT and lower THI at end-surgery as compared to that observed in the balanced anaesthesia group. Balanced anaesthesia was associated with an increase in microvascular reactivity, as reflected by a greater area of hyperaemia after the VOT as compared to that seen at baseline.

The activation of the inflammatory response may occur early during aortic surgery [33] with maximal production of oxygen free radicals within 5 min of lower-limb reperfusion [34]. Previous studies showed that oxidative stress during supraceliac or infrarenal aortic cross clamping and reperfusion leads to distributive alterations in microvascular oxygenation and perfusion of splanchnic organs [35]. Heterogeneous flow distribution is characterized by capillaries with stagnant blood cells next to capillaries with normal flow [36] and can be found in microvascular beds far from the site of direct ischemic damage [37]. In rats subjected to I/R injury of the small bowel, the intestinal microvascular blood flow was reduced early in the first 10 min of reperfusion and failed to normalize in the first 2 h [38]. Microcirculatory disturbances during I/R injury may occur independently of changes in macro-haemodynamics and tissue hypo-perfusion may persist despite a normalisation of systemic cardiocirculatory parameters due to a loss of haemodynamic coherence [39]. In our study, we did not find an impairment in microvascular perfusion at end-surgery. We hypothesize that intraoperative hemodynamic optimization with the routine use of a protocol of goal-directed therapy prevented the development of microvascular derangements. In pigs subjected to colon anastomosis, a goal-directed colloid administration was able to improve microvascular perfusion and oxygenation of healthy and perianastomotic colon [40]. Although we could not find any correlation with MAP or CI, changes in sublingual capillary density were inversely related with changes in systemic vascular resistance, reflecting a certain degree of coherence between the macro- and micro-haemodynamic responses in our population [41].

Sublingual microvascular density increased at end-surgery in patients receiving balanced anaesthesia, while it remained unaltered in the TIVA group. It cannot be excluded that the microvascular changes observed were the effect of vasodilating metabolites (e.g. adenosine monophosphate, hypoxanthine) released during reperfusion, however we would expect this reactive hyperaemia to disappear in a late reperfusion phase at end-surgery [3]. Even if the observational design of our study prevents to define cause-effect relationships, the between-group homogeneity in baseline and intraoperative characteristics (including clamping time, fluid balance, blood losses, type and dose of vasoactive agents) would corroborate the hypothesis of direct and separate actions of desflurane and propofol on the microcirculation.

Previous studies showed that both propofol and sevoflurane can exert a protective effect towards I/R injury by modulating the inflammatory response, reducing oxidative stress and tissue apoptosis [42]. Halogenated anaesthetics induce peripheral vasodilation [43] and reduce vascular permeability [23], thus favouring microvascular recruitment and tissue oxygen diffusion. On the other hand, propofol can modulate the inflammatory response and oxidative stress [44], thus increasing tolerance to hypoxia also in tissues far from those directly exposed to ischemia [45]. In this study, desflurane may have been responsible for vasodilation and vascular recruitment in the balanced anaesthesia group due to dilation of resistive arterioles. The increased area of hyperaemia after the VOT would also be consistent with a greater microvascular recruitment following the ischemic challenge in the balanced anaesthesia group.

The rationale of evaluating the microcirculatory response to therapies is that improving microvascular perfusion will guarantee tissue oxygenation and prevent the development of organ dysfunction. In a previous study in patients undergoing abdominal aortic aneurysm surgery, we showed that portal lactate, intramucosal sigmoid pH and arterial-sigmoid delta pCO₂ (as indices of splanchnic hypo-perfusion) were able to predict the occurrence of post-operative complications [46]. In the present study, we could not show a better impact on anaerobic metabolism of desflurane or propofol, as a clinically relevant hyper-lactataemia was not found in either group at end-surgery. This was consistent with the absence of major microcirculatory alterations in both the balanced anaesthesia and the TIVA groups.

While the StO₂ remained stable in both groups, the desaturation rate in the first part of the ischemic challenge decreased at end-surgery in patients undergoing TIVA. This may result from a reduction in peripheral microvascular density and would be consistent with the lower THI at end-surgery as compared to that seen in the balanced anaesthesia group, which seems to be related to a lower tissue perfusion rather than to a reduction in Hb levels as these remained similar in the two groups. Baseline differences in StO₂ downslope between the groups, although not significant, could also explain our results: the observed variation may only depend upon a rebalance in the metabolic pattern in the TIVA group without a dysfunctional meaning. In critically ill patients, we showed that a decrease in the desaturation rate in the last part of the ischaemic phase of the VOT is associated with worse outcome and could suggest impaired microcirculatory auto-regulation [32]. In this study, the StO₂ downslope was again similar between the groups in the final part of vascular occlusion and the delta-downslope did not differ between the two groups. Therefore, the type of anaesthesia may not significantly affect peripheral microcirculatory auto-regulation.

In patients undergoing open abdominal aortic aneurysm repair under general anaesthesia with a protocol of intraoperative hemodynamic goal-directed therapy, microvascular perfusion and peripheral tissue oxygenation were generally preserved, however the use of balanced anaesthesia was associated with increased microvascular density and reactivity, while these remained unaltered with TIVA. Further studies are needed to clarify whether the choice of different anaesthetic regimens may contribute to prevent I/R injury in aortic surgery.