Background
Extracorporeal Circulation (ECC) is associated with significant changes in the physiology of peripheral perfusion [
1,
2]. The main mechanisms by which ECC leads to an impairment in organ perfusion are the activation of inflammatory pathways with consequent endothelial damage, capillary leak [
3], interstitial oedema, and impaired microcirculatory blood flow. Consequently, even when global O
2 delivery is preserved, a local hypoxia occurs, leading to organ dysfunction that is associated with worse patient outcomes [
4]. Near infrared spectroscopy (NIRS) is a non-invasive technique that allows static and dynamic assessments of tissue oxygen saturation in response to ischemic challenge (vascular occlusion test, VOT), thus providing information of peripheral (skeletal muscle) O
2 extraction rate and microvascular reactivity [
5,
6]. Alterations in tissue oxygen saturation are associated with higher mortality in patients with sepsis [
7]. NIRS monitoring can provide information on the effects that therapies may have on tissue perfusion and microcirculation [
8]. Only a few single-centre studies investigated the relationship between peripheral NIRS-derived parameters and patients’ outcome in cardiac surgery, showing conflicting results and using different devices to assess tissue oxygen saturation [
9‐
15]. Postoperative complications and persistent elevated arterial lactate concentrations were associated with low StO
2 after ICU admission [
9,
12]. By using a vascular occlusion test (VOT), some studies showed that alterations in the desaturation and reperfusion slopes in the early post-operative phase following cardiac surgery were associated with poor outcome, duration of mechanical ventilation, length of ICU stay, and mortality [
11,
13,
15]. Conversely, other studies failed to demonstrate the association between NIRS-derived parameters and outcome [
10,
14].
The aim of the present bicentric study was to assess the ability of NIRS-derived parameters to predict postoperative complications in patients undergoing cardiac surgery with ECC.
Discussion
In the present study of 90 patients undergoing cardiac surgery with ECC, NIRS monitoring at the thenar eminence with VOT detected a significant reduction in skeletal muscle microvascular reactivity following surgery, together with a decrease in tissue O2 extraction rate, which did not recover in the first 6 h after the operation. We failed to detect an association between NIRS-derived parameters and patient outcome since patients with post-operative complications showed similar variations as those without complications.
It is well known that cardiac surgery with ECC can induce a complex inflammatory response. This can be due to multiple factors, including surgical trauma, haemodilution, ischemia/reperfusion injury, hypothermia and exposure of blood to non-physiological surfaces [
1,
3]. The mechanisms involve the following: the release of cytokines, complement activation, leukocyte activation with endothelial adhesion, an increased production of O
2 free radicals, the release of inflammatory mediators including endothelin, and the deregulation of the nitric oxide pathway [
24]. Under these conditions, systemic haemodynamic parameters and markers of global oxygenation, such as central venous O
2 saturation or arterial lactate, may not be early predictors of tissue hypoperfusion. Although increased lactate levels are related to morbidity and mortality in different patient groups [
25], they lack sensitivity and specificity in representing tissue perfusion and may not be sufficient to detect the early impairment of tissue oxygenation [
26‐
28]. The pathological mechanism triggered by the ECC showed strong similarities to those seen during sepsis, potentially leading to impaired microcirculatory perfusion and tissue hypoxia. Studies using sublingual videomicroscopy have shown alterations in microvascular perfusion, which may persist for 24 h after surgery [
29] and may occur irrespective of changes in systemic haemodynamics [
30]. Bauer et al. showed an increased number of rolling leukocytes in the sublingual microcirculation during CPB, which persisted 1 h after the termination of CPB [
31].
NIRS monitoring in conjunction with VOT enables us to estimate peripheral tissue oxygen saturation and microvascular reactivity by evaluating variations in StO
2 during a brief ischemia/reperfusion test [
5]. In several studies using this technology, patients undergoing cardiac surgery showed impaired microvascular reactivity, although conflicting data exists regarding the time needed for recovery to the baseline microvascular state and the relationship with outcome. Smith et al. showed that during CPB, the reperfusion slope decreased as a function of CPB duration, returning to baseline values in all patients within 1 h of the termination of CPB [
32]. In another study by Morel et al., StO
2 and reperfusion slope both declined after CPB but recovered to baseline values after 12 h [
10]. Furthermore, these transient changes were not correlated with the patients’ outcome. We found a similar trend in the recovery slope at the first 6 h of the study; however, we did not collect NIRS parameters after 6 h. There are two main differences between our study and that by Morel et al. [
10]. Firstly, we used major complications as variables of outcome, instead of using the Sequential Organ Failure Assessment score. Secondly, the duration of VOT was different in the two studies: we applied a three minute time targeted VOT, while Morel et al. maintained the VOT until the StO
2 value reached 40%. However, despite these differences, our results were consistent with those showed by Morel [
10]. Kim et al. demonstrated that the reperfusion slope largely recovered on the first day after surgery in patients without complications, while it remained altered in those with complications [
12]. In the present study, the recovery slope remained reduced in patients for 6 h after admission to the ICU, indicating a persistent decrease in microvascular reactivity. However, we were unable to detect more severe alterations or delayed recovery of this parameter among patients with post-operative complications in such a short monitoring period. There may be several explanations for these discrepant results. First, we performed 3-min blood flow occlusion in all patients, instead of using a target StO
2, and this may have produced different degrees of ischemia. It has been shown that StO
2 recovery rate depends on the minimum StO
2 reached after 3 min of blood flow occlusion; i.e. the velocity of reperfusion and the degree of hyperaemia are related to the degree of ischemia [
33]. Second, it is possible that the monitoring period in this study was too short to detect differences for predicting post-operative morbidity.
Kopp et al. showed that StO
2 was reduced after cardiac surgery and that the minimum value of StO
2 in the first hours after the operation was predictive of delayed lactate clearance [
13]. However, in a similar patient population other authors have found a transient increase in StO
2 after surgery [
34]. StO
2 reflects the balance between regional O
2 delivery and consumption [
5]. In our study, in the first 6 h after surgery, we observed an increase in StO
2, a slower desaturation rate (flatter occlusion slope), smaller area of ischemia, and increased StO
2 values (nadir) during 3-min of ischemia, but no difference was seen between patients with and without complications. Taken together, these findings suggest a reduction in regional O
2 consumption, despite systemic O
2ER resulting in the normal range in both groups of patients. Several factors may influence O
2 extraction and consumption in skeletal muscle, including the administration of sedative or vasopressor agents, a residual neuromuscular blockade in the first post-operative hours, and variations in body temperature [
4,
35].
Our study has several limitations. First of all, we studied a convenience sample of 90 patients without preliminary statistical calculation of the required sample size. Based on previous studies, we retrospectively calculated that the inclusion of a total of 96 patients was required to show a difference in the reperfusion slope between patients with, and those without, post-operative complications (calculated Cohen’s d = 0.58 [
12]) with a power of 80% and an alpha error of 0.05. Therefore, our study may be slightly underpowered, although it is highly unlikely that the inclusion of 6 additional patients would have changed our results substantially. Second, by applying NIRS to the thenar eminence, we evaluated tissue oxygen saturation and microvascular reactivity in a peripheral tissue. We cannot determine whether similar alterations were induced in microvascular beds of the splanchnic organs. We used the thenar eminence because the thickness of the adipose tissue covering this muscle is small. Although tissue oedema can increase the thickness of the subcutaneous layer, this hardly happens in a period of time as short as that of our study [
20]. Third, since the degree of neuromuscular blockade was not measured in our patients, it is possible that the reduction in muscle O
2 extraction rate may at least partly depend on a residual neuromuscular blockade in the early postoperative period. Moreover, the use of vasopressors, inotropes and transfusions, as well as the perioperative fluid balance, could affect peripheral tissue oxygen supply and utilization. Unfortunately however, we could not evaluate the potential influence of these treatments because data on vasoactive dosage or fluid and transfusion requirements were not collected. Similarly, we were unable to evaluate the impact of different ECC protocols (e.g the use of roller versus centrifugal pump). Fourth, we did not perform intraoperative NIRS measurements that could provide information about earlier variations in tissue oxygen saturation and microvascular function and their impact on clinical outcomes. Fifth, a target StO
2 for VOT, instead of a pre-defined occlusion time, may have been more appropriate for standardizing the degree of ischemia and the hyperaemic phase [
33]. Furthermore, we did not collect the tissue haemoglobin index, usually used to calculate the muscle oxygen consumption (NirVO
2); therefore, we could not evaluate the impact of variations of the NirVO
2 on the outcome. Sixth, we did not evaluate the potential role of intraoperative complications or risky events (e.g., arterial hypotension, haemorrhage, low cardiac output) on NIRS parameter and their relationship with the outcome. Again, the tissue spectrometer (InSpectra Model 650) used in the present study, and the company that produces it (Hutchinson Technology, Hutchinson, MN, USA) are actually out of the market. Similar devices using near infrared spectroscopy exist that can provide tissue O
2 saturation. Nonetheless, it must be recognized that differences in the proprietary algorithms of different NIRS devices make the comparisons between studies difficult.
Finally, we performed multiple comparisons on a number of variables, thus our analysis may be affected by bias due to multiple-testing and data coupling problems. Indeed, the slopes for the decreasing and increasing StO2 during the VOT are not independent from the areas that are defined by the slopes; therefore, a data coupling may occur. Nonetheless, we applied the Bonferroni correction to enhance the robustness of our results, and NIRS variables were included individually in separate logistic regression models in order to avoid the problem of collinearity. Even if the study is observational, not having registered it in any public register, the designs and statistical analyses are unverifiable.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.