Baseline cerebral oximetry values in cardiac and vascular surgery patients: a prospective observational study

Aim This study was conducted to evaluate baseline INVOS values and identify factors influencing preoperative baseline INVOS values in carotid endarterectomy and cardiac surgery patients. Methods This is a prospective observational study on 157 patients (100 cardiac surgery patients, 57 carotid endarterectomy patients). Data were collected on factors potentially related to baseline INVOS values. Data were analyzed with student's t-test, Chi-square, Pearson's correlation or Linear Regression as appropriate. Results 100 cardiac surgery patients and 57 carotid surgery patients enrolled. Compared to cardiac surgery, carotid endarterectomy patients were older (71.05 ± 8.69 vs. 65.72 ± 11.04, P < 0.001), with higher baseline INVOS (P < 0.007) and greater stroke frequency (P < 0.002). Diabetes and high cholesterol were more common in cardiac surgery patients. Right side INVOS values were strongly correlated with left-side values in carotid (r = 0.772, P < 0.0001) and cardiac surgery patients (r = 0.697, P < 0.0001). Diabetes and high cholesterol were associated with significantly (P < 0.001) lower INVOS and smoking was associated with higher INVOS values in carotid, but not in cardiac surgery patients. Age, sex, CVA history, Hypertension, CAD, Asthma, carotid stenosis side and surgery side were not related to INVOS. Multivariate analysis showed that diabetes is strongly associated with lower baseline INVOS values bilaterally (P < 0.001) and explained 36.4% of observed baseline INVOS variability in carotid (but not cardiac) surgery. Conclusion Compared to cardiac surgery, carotid endarterectomy patients are older, with higher baseline INVOS values and greater stroke frequency. Diabetes and high cholesterol are associated with lower baseline INVOS values in carotid surgery. Right and left side INVOS values are strongly correlated in both patient groups.


Introduction
Persistent cognitive decline or permanent neurologic deficits are common after cardiac or vascular surgery [1]. A large prospective study reported that serious neurological deficits occur in up to 6.2% of patients after myocardial re-vascularization [2], and factors other than emboli seem to be involved in more than 50% of cases. A study by Slater et al [3] showed that the incidence of early postoperative cognitive decline was 60%. Other data show that more than 40% of patients undergoing cardiac surgery develop persistent cognitive decline resulting in functional impairment [4] and prolonged hospital stay [3], and, according to current thinking, embolism is not the sole cause of these phenomena. Cerebral oximetry, as measured by INVOS, is a promising neuro-monitoring technology [5], but its usefulness during cardiac surgery, vascular surgery, and in the cardiovascular ICU has not, as of yet, been adequately evaluated.
Non-invasive cerebral oximetry uses near-infrared reflectance spectroscopy (NIRS) to measure frontal lobe regional cortical oxygen saturation. Measurement is based on the different absorption characteristics of oxygenated and deoxygenated hemoglobin: oxygenated hemoglobin (HbO 2 ) absorbs less red light (600-750 nm) and more infrared light (850-1000 nm) than deoxygenated hemoglobin. As a result, deoxygenated hemoglobin has an absorption peak at 740 nm while HbO 2 does not [5]. Consequently, the fraction of oxyhemoglobin can be determined by using two infrared wavelengths, thereby providing a technique for continuous non-invasive, bed-side monitoring that reflects the balance between cerebral oxygen supply and demand [5]. Other techniques, such as jugular venous saturation and electroencephalography have also been used [6], but in this study we only evaluated INVOS.
An association between cerebral oxygen desaturation during cardiac surgery and postoperative cognitive dysfunction, prolonged intensive care unit (ICU), and hospital stay has been demonstrated [7], and intraoperative cerebral ischemia and cerebral oxygen desaturation have been proposed as possible mechanisms contributing to postoperative cognitive dysfunction [7,8]. In addition, a RCT conducted by Murkin and colleagues [9] demonstrated that treatment of cerebral oxygen desaturation was associated with shorter ICU length of stay, significantly reduced incidence of major organ morbidity, and lower mortality. Cerebral oximetry monitoring is increasingly used to monitor frontal lobe perfusion during cardiac and non-cardiac surgery. Furthermore, the use of INVOS has been reported to help detect aortic cannula displacement, and some authors have suggested that all cardiac surgery patients should have intraoperative cerebral oxygenation monitoring [10].
Perioperative stroke is an inherent risk of carotid endarterectomy and occurs in 5-7.5% of patients [11]. As hypoperfusion during cross clamping is a major cause of stroke, CEA can be considered as a human model of regional cerebral ischemia, and may provide an ideal opportunity for evaluating the role of INVOS as a monitor of cerebral ischemia.
Not surprisingly, cerebral oximetry has been used in several investigations on patients undergoing CEA [5], and there is significant correlation between carotid stump pressure and cerebral oximetry during carotid endarterectomy [12]. In the last decade, technological research has expanded the application of NIRS to allow continuous, non-invasive bed-side monitoring of cerebral tissue oxygen saturation through the scalp and skull, thereby providing accurate useful information on the balance between brain oxygen supply and demand [5]. Due to the variability of baseline rSO 2 values between patients, a baseline should be determined for each patient before induction of general anesthesia, and detection of cerebral ischemia is based on deviations from baseline, rather than on absolute INVOS values. Generally, a 20% reduction below baseline is considered evidence of cerebral ischemia [13,14]. However, if baseline rSO 2 is < 50%, then reduction by 15% below baseline is the critical threshold for ischemia detection. Data suggest that routine use of rSO 2 monitoring to guide the anesthesia plan during cardiac surgery may improve patient outcome and shorten hospital stay [5,11,15]. Several studies have attempted to define the risk factors and the conditions influencing rSO 2 baseline, and age is considered the strongest predictive factor for postoperative cognitive dysfunction (POCD) after cardiac surgery [16]. In addition to advanced age, other reported risk factors for POCD after coronary artery bypass graft surgery (CABG) are systemic inflammation [17], low education level, diabetes, severity of atherosclerotic disease and type of surgery [1,16].
This study was conducted to determine factors associated with preoperative baseline INVOS values in patients undergoing CABG, valve replacement or carotid endarterectomy surgery. Hematocrit, sex, anthropometric characteristics, blood oxygenation, cerebral blood flow, cerebral metabolic rate and head position can influence rSO 2 [5]. Hypocarbia, and inadequate mean arterial pressure (MAP) are additional factors influencing rSO 2 [18]. In this study we attempted to evaluate the relationship, if any, of other variables, such as left ventricle ejection fraction, side of carotid stenosis, history of cardiac ischemic and/or cerebrovascular event on baseline preoperative INVOS values.

Methods
This prospective, non-randomized, observational study was conducted at the University Hospital of Ioannina between October 2007 and December 2008. The study was approved by the Institution Ethics Committee, and all patients gave written informed consent for data collection. 100 patients undergoing cardiac surgery and 57 patients undergoing carotid surgery enrolled.
Inclusion criteria were elective carotid or cardiac surgery and age > 18.
Exclusion criteria were: emergency surgery, surgery starting after 18.00, age > 90, renal failure requiring hemodialysis, advanced liver cirrhosis with elevated baseline bilirubin or prolonged PT, known dementia and known serious psychiatric disease.
Fifty seven patients scheduled for elective carotid endarterectomy, and 100 patients scheduled for elective cardiac surgery with or without cardiopulmonary bypass (CPB) enrolled. All carotid endarterectomy operations were performed by the same vascular surgeon (MM) without using a shunt. Likewise, all cardiac operations were performed by the same cardiac surgeon (SS). Among patients undergoing cardiac surgery (n = 100), 78 patients had CABG (42 patients with CPB, 36 patients without CPB) and 22 patients had valve replacement surgery.
Demographic data and data on risk factors known or believed to be associated with coronary artery and/or peripheral vascular disease (Age, Gender, Diabetes Mellitus, History of Stroke, Smoking, High cholesterol, Hypertension) were prospectively collected from all patients. Right and Left side baseline INVOS data were recorded before oxygen administration started and before any sedation was given.

Data collection and analysis
As this is an observational study, we did not conduct any power analysis for sample size estimation, and there was no randomization or blinding. Data were prospectively collected and securely stored in an electronic database.
All data analysis was done with the SPSS v. 16 statistical software package (SPSS Inc, Chicago, IL). Data normality was assessed with the Kolmogorov Smirnov test. Depending on data distribution, continuous data were compared with two-sided Student's t test or the Mann-Whitney U test. Correlations between variables with continuous data were assessed with Pearson's r, and comparisons between proportions were done with Chisquare test. P < 0.05 was considered significant for all comparisons. Linear regression was used to analyze the relative contribution of different variables to observed baseline INVOS variability. The "Statistica" version 7 Statistical Software Package (StatSoft Inc, Tulsa, Oklahoma, USA) was used to generate scatter plots for significant correlations between variables.

Results
A total of 157 patients enrolled; 100 of those had cardiac surgery and 57 had carotid surgery. Demographic data, risk factors and baseline preoperative INVOS values are presented in Table 1. Patients undergoing carotid surgery were significantly older, and had higher baseline INVOS values and greater frequency of stroke. Diabetes and high cholesterol were significantly more common among cardiac surgery patients (Table 1).

INVOS in vascular surgery
Baseline INVOS values in vascular surgery patients had normal distribution bilaterally. Comparison between the right-sided (Table 2) and left-sided (Table 3)  Diabetes, smoking and high cholesterol were associated with cerebral oximetry: baseline INVOS values were significantly lower bilaterally in patients with DM (60.08 ± 9.03 on the left, 57.00 ± 6.90 on the right) compared to patients who did not have DM (68.80 ± 6.82 on the left, 68.55 ± 6.34 on the right, P < 0.000).
Baseline INVOS values were also related to smoking, with smokers having higher INVOS values on the left (68.20 ± 7.03 vs. 63.25 ± 9.94 in non-smokers, P < 0.039). Age, sex, history of CVA, Hypertension, Presence of CAD, Presence of Asthma, Side of carotid stenosis and Side of carotid surgery (Table 4) were not related to INVOS values. The relationship between the above variables and baseline INVOS values was evaluated with Multivariate analysis, which also showed that Diabetes is significantly associated with lower baseline INVOS    (Table 5) and left side ( Table 6) are strongly correlated (r = 0.697, P < 0.0001). Correlation between right and left INVOS values is shown graphically in Figure 2.
In contrast to our findings in carotid surgery patients, diabetes, smoking and high cholesterol were not associated with baseline cerebral oximetry values in cardiac surgery patients. Age, gender, history of old MI, Hypertension, and the type of operation (valve replacement vs. CABG) were not related to baseline INVOS values on either side.
Linear regression analysis was used to search for variables that could predict right or left-sided baseline INVOS values. Regression was done on 92 cases (8 cases contained missing values), and showed that LVEF and baseline right-side baseline INVOS values are independent, significant predictors of left-side INVOS values. In addition to regression, we also looked for correlations between baseline R or L side INVOS values and weight, height, LVEF and Euroscore. This analysis showed that L-sided INVOS is marginally correlated with body weight (r = 0.192, p < 0.061) and significantly correlated with LVEF (r = 0.206, p < 0.043, Figure 3), whereas the correlation between L-sided INVOS and Euroscore was negative, but did not reach statistical significance (P = 0.09). In contrast, the correlation between R-sided INVOS and Euroscore was negative and significant (r = -0.315, p < 0.001, Figure 4).        As this is an observational study, and there was no intervention in response to observed INVOS values, our data cannot support any conclusions regarding intraoperative management of these patients. However, this prospective observational study provides some direction for future research on factors that may influence baseline and intraoperative INVOS values, but our patient number is relatively small, and does not allow for definite conclusions. Data from larger prospective studies are needed to evaluate the validity of our findings.