Incidence of Concurrent Cerebral Desaturation and Electroencephalographic Burst Suppression in Cardiac Surgery Patients

Abstract

Background

Increased intraoperative electroencephalographic (EEG) burst suppression is associated with postoperative delirium. Cerebral desaturation is considered as one of the factors associated with burst suppression. Our study investigates the association between cerebral desaturation and burst suppression by analyzing their concurrence. Additionally, we aim to examine their association with cardiac surgical phases in order to identify potential for targeted interventions.

Methods

We retrospectively analyzed intraoperative 1-minute interval observations in 51 patients undergoing cardiac surgery. Processed EEG and cerebral oximetry were collected, with the anesthesiologists blinded to the information. The associations between cerebral desaturation (defined as a 10% decrease from baseline) and burst suppression, as well as with phase of cardiac surgery, were analyzed using the Generalized Logistic Mixed Effect Model. The results were presented as odds ratio and 95% confidence intervals (CI). P<0.05 was considered statistically significant.

Results

The odds of burst suppression increased 1.5 times with cerebral desaturation (OR 1.52, 95% CI [1.11–2.07], p=0.009). Compared to pre-cardiopulmonary bypass (pre-CPB), the odds of cerebral desaturation were notably higher during CPB (22.1, 95% CI [12.4–39.2]; p<0.001) and post-CPB (18.2, 95% CI [12.2–27.3]; p<0.001). However, the odds of burst suppression were lower during post-CPB (0.69, 95% CI [0.59, 0.81]; p<0.001) compared to pre-CPB. Compared to pre-CPB, the odds of concurrent cerebral desaturation and burst suppression were notably higher during CPB (52.3, 95% CI [19.5–140]; p<0.001) and post-CPB (12.7, 95% CI [6.39–25.2]; p<0.001). During CPB, the odds of cerebral desaturation (6.59, 95% CI [3.62–12] p<0.001) and concurrent cerebral desaturation and burst suppression (10, 95% CI [4.01– 25.1], p<0.001) were higher in the period between removal of aortic cross-clamp and end of CPB. During the entire surgery, the odds of burst suppression increased 8 times with higher inhalational anesthesia concentration (OR: 7.81, 95% CI [6.26 – 9.74], p<0.001 per 0.1% increase).

Conclusion

Cerebral desaturation is associated with intraoperative burst suppression during cardiac surgery, most significantly during CPB, especially during the period between the removal of the aortic cross-clamp and end of CPB. Further exploration with simultaneous cerebral oximetry and EEG monitoring is required to determine the causes of burst suppression. Targeted interventions to address cerebral desaturation may assist in mitigating burst suppression and consequently enhance postoperative cognitive function.

Summary statement:

Cerebral desaturations and burst suppressions occurred frequently. Concurrent cerebral desaturations and burst suppression were most frequent during the period between removal of aortic cross-clamp and end of CPB.

Introduction

Perioperative neurocognitive disorders are common after cardiac surgery and have a significant short- and long-term impact on health and well-being^1,2. Intraoperative electroencephalography (EEG) burst suppression, defined by periods of marked suppression alternating with bursts of brain electrical activity predicts postoperative delirium³. Implementing intraoperative algorithms that utilize perioperative EEG monitoring for anesthetic depth management could possibly reduce the incidence of neurocognitive-related morbidity ^4,5.

EEG burst suppression, typically attributed to increased anesthetic depth⁶, can also be caused by cerebral hypo-perfusion due to intraoperative hypotension, anemia or embolic phenomena⁷. Cerebral hypo-perfusion has been associated with postoperative delirium⁸. In a clinical case series, an algorithm incorporating the Patient State Index (PSI) and cerebral oximetry was proposed for patient management during cardiac surgery⁹. The PSI is a clinically validated measure of the effect of anaesthesia and sedation. The PSI is calculated by a high-resolution 4-channel electroencephalograph (EEG) monitor via a proprietary algorithm after advanced artifact rejection (Masimo Inc.. Irvine, California) ¹⁰. In that series, a low PSI with concurrent cerebral desaturation was attributed to cerebral hypo-perfusion. Similar clinical algorithms have been implemented, aiming to reduce postoperative cognitive decline following non-cardiac surgeries^11,12.

Previous literature has shown that cardiac surgical patients are at increased risk of neurological events during cardiopulmonary bypass (CPB), especially towards its end, when the aortic cross-clamp is being removed. This susceptibility was assessed through the detection of cerebral emboli and monitoring of cerebral metabolism^13,14.

The incidence of concurrent cerebral desaturation and burst suppression, both of which may be associated with significant cerebral hypo-perfusion, during the various phases of cardiac surgery remains to be investigated. We examined the incidences of cerebral desaturation, burst suppression, and their concurrency during different periods of cardiac surgery.

Materials and methods

The study population included patients from an ongoing multisite randomized controlled trial (PANDORA), which was designed to evaluate the effect of postoperative intravenous acetaminophen vs. placebo to prevent postoperative delirium¹⁵. This trial is approved and centrally facilitated by the Institutional Review Board of Beth Israel Deaconess Medical Center – BIDMC (Protocol #2019P000758). The trial was registered before patient enrollment at ClinicalTrials.gov (NCT04093219, Principal Investigator: Balachundhar Subramaniam, date of registration: September 13, 2019). All patients provided written or electronic informed consent. The study participants are cardiac surgical patients aged >60 years requiring full CPB undergoing coronary artery bypass grafts and/or valve repair. Aortic surgeries were excluded. As part of the trial, delirium assessments were performed using the Confusion Assessment Method (CAM) at baseline and postoperatively until discharge¹⁶.

This study is a retrospective analysis of prospectively collected blinded EEG and cerebral oximetry data from a single site (BIDMC). Both EEG and cerebral oximetry monitoring were started in the preoperative waiting room with the application of an EEG and a NIRS sensor on the participants' foreheads, connected to a single monitor; the data were continuously collected till the end of surgery. Cerebral oximetry was measured using near-infrared spectroscopy (NIRS) [Masimo Inc., Irvine, California]. The EEG sensor has four leads, providing four simultaneous channels of frontal EEG waveforms. SEDline monitors and sensors (Masimo Inc., Irvine, California) were used to collect EEG data. The clinicians were blinded to the data and data were not used for patient management during surgery. Typically, at our institution during the study period, anesthesia was induced with intravenous fentanyl and propofol or etomidate (based on patient characteristics and anesthesiologist preference), and rocuronium was given for skeletal muscle relaxation. Anesthesia was maintained with isoflurane 0.5%–1.0% in 100% oxygen with supplemental intravenous fentanyl given as intraoperative analgesia. Patients were ventilated with a tidal volume of 6 ml/kg, and respiratory rate was adjusted to maintain a PCO2 of 40–55 mmHg and a pH higher than 7.25. Patients were placed on a non-pulsatile CPB pump using arterial and venous cannulae. The perfusionist titrated CPB flow to achieve a venous blood saturation of >60% and a mean arterial pressure (MAP) of 50–70 mm Hg unless otherwise specified in discussion with the attending anesthesiologist and surgeon.

Burst Suppression Ratios (SR) were automatically calculated by the monitor as the percentage of EEG recording showing burst suppressions during the previous minute, refreshing every 2 seconds¹⁷. The values were registered as whole integers from 0–100%. An observation showing SR of 1% indicates 0.6 seconds (1% of a minute) of burst suppression in the last minute and is the lowest discrete incidence of burst suppression captured. Regional cerebral oximetry measurements (average from both sides) were recorded as real-time observations (refreshed every 2 seconds) from the right and left cerebral hemispheres. Retrospectively, data was downloaded from the monitor for research purposes. The SR values were compared with the cerebral saturation averaged over the previous minute for analysis of their concurrence. Concurrent cerebral desaturation and burst suppression were defined by simultaneous observations of SR>0 and cerebral desaturation (defined as a 10% decrease from baseline)^18,19. Baseline values for cerebral oximetry were set in the preoperative waiting room. To study the association between the phases of surgery, data was divided into pre-CPB, CPB, and post-CPB periods. The CPB period was further divided into the period between removal of aortic cross-clamp and end of CPB. Baseline demographic parameters were collected from medical records. Hemodynamic and anesthetic data were collected from the anesthesia information system. This manuscript follows the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guidelines.

Statistical analysis

A priori sample size calculation was not conducted and a convenience sample was utilized. We analyzed the associations for three outcomes: i) burst suppression; ii) cerebral desaturation; and iii) concurrent cerebral desaturation and burst suppression. We further analyzed the associations within CPB in separate models. Robust statistical distribution tests for normality were done for the outcomes, before the analyses. To account for the non-independent nature of the variables, we used Generalized Linear Mixed mixed-effect models with a logit link function. Specifically, the phases of surgery (pre-CPB, CPB, and post-CPB) and the phases of CPB (the phase reflective of the period between the aortic cross-clamp removal and the end of CPB and the remaining of CPB, were modeled as categorical variables in separate models. The models used random intercept of individual patient identifiers (study IDs) adjusting for the covariates age, gender, type of surgery, end-tidal carbon dioxide (etCO2), and isoflurane concentration. This approach allowed us to effectively address the correlated or dependent nature of observations. The etCO2 levels and isoflurane concentrations were collected as per-minute observations and aligned with other variables based on time for the analyses. The results are presented as odds ratios with 95% confidence intervals (CI). All statistical analyses were performed using R version 4.2.2. P<0.05 was considered statistically significant.

Results

Enrollment of participants in the PANDORA clinical trial began in 2020. Baseline characteristics are presented in Table 1. None of the study participants showed delirium on baseline. Of 117 participants in the study center, 51 had data available for simultaneous cerebral oximetry and burst suppression observations for the entire surgery. Mean (Standard Deviation, SD) baseline cerebral oximetry values from the left and right sides were 63% [ (9) and 63% (SD 9) (Table 1), respectively. 11,878 one-minute interval observations (with data for both cerebral oximetry and burst suppression) were obtained from 51 participants. The median number of observations contributed by each participant was 250 (IQR: 215, 304). The median [IQR] duration of pre-CPB, CPB, and post-CPB periods were 115 [93, 140] minutes, 87.5 [66.5, 97.5] minutes, and 56 [43, 67.5] minutes, respectively.

Table 1.

Demographics and baseline characteristics of patients included in the study

Characteristic	N = 51
Age [years]	70 [66.1, 73.9]
Weight [kg]	83 [73, 95]
Body Mass Index [Kg/m 2 ]	27.8 [24.7, 31.5]
Female	12 (24%)
Race
Black or African-American	1 (2.0%)
White	49 (96%)
Other	1 (2.0%)
Cardiac Surgery Type:
CABG with or without valve surgery	34 (67%)
Valve surgeries alone	17 (33%)
Cerebral oximetry (Left)	63 [59, 68]
Cerebral oximetry (Right)	62 [59, 68]

Values are represented as Median [IQR]; Number of participants are represented as percentages

Figure 1 shows the incidence of cerebral desaturation, burst suppression and their concurrency. Among the 11,878 one-minute interval observations, the incidence of burst suppression was 22% and cerebral desaturation 14%. Within burst suppression, 3% were concurrent with cerebral desaturation and 13% with isoflurane concentrations >1.5%. Figure 2 and Supplemental Table 1 show the summary statistics of cerebral desaturation durations during pre-CPB, CPB, and post-CPB phases. The median duration between aortic cross-clamp removal and end of CPB averaged 10.5 minutes. Consequently, we approximated this period as the last 10 minutes of CPB. Supplemental Figure 1 and 2 shows the changes in cerebral oximetry from baseline for both the left and right sides.

Figure 1a.

Incidence of cerebral desaturation, burst suppression, and concurrent events before, during, and after cardiopulmonary bypass

Median percentages (95% Confidence Intervals)

Percentages were calculated with the total number of observations during each phase of the surgery as the denominator

Figure 2.

Summary statistics of cerebral desaturation (>10 % decrease from baseline) durations during pre-cardiopulmonary bypass (pre-CPB), CPB, and post-CPB phases

Median durations (IQR)

Associations for burst suppression (Table 2)

Table 2.

Factor associated with intraoperative burst suppression

Characteristic	OR	95% CI1	p-value	p-value (Type III)*
Cerebral desaturation [vs normal cerebral oximetry]	1.52	1.11, 2.07	0.009
CPB [vs pre-CPB]	1.27	0.95, 1.69	0.11	< 0.001
Post-CPB [vs pre-CPB]	0.69	0.59, 0.81	<0.001
Age	1.1	1.02, 1.18	0.009
Female gender [vs male)]	1.82	0.67, 4.99	0.2
Valve surgeries alone [vs CABG +/− valve]	0.72	0.29, 1.78	0.5
EtCO2	0.99	0.98, 0.99	<0.001
Isoflurane concentration	7.81	6.26, 9.74	<0.001

¹CI = Confidence Interval

^*p-values for categorical variables of more than two categories are shown

Burst suppression was more often associated with cerebral desaturation (1.52, [1.11– 2.07], p=0.009) than normal cerebral saturation, after adjusting for covariates. Compared to pre-CPB, the odds of burst suppression were reduced post-CPB (0.69 [0.59– 0.81], p<0.001). The odds of burst suppression did not significantly differ between CPB and pre-CPB (1.27 [0.95–1.69], p=0.11). An increase in isoflurane concentration (per 0.1%) was associated with higher odds of burst suppression : 7.81, [8.45– 13.6], p<0.001.

Associations for cerebral desaturation (Table 3)

Table 3.

Factors associated with cerebral desaturation

Characteristic	OR	95% CI1	p-value	p-value* (Type III)
Burst suppression [vs no burst suppression]	1.16	0.83, 1.63	0.4
CPB [vs pre-CPB]	22.1	12.4, 39.2	<0.001	< 0.001
Post CPB [vs pre-CPB]	18.2	12.2, 27.3	<0.001
Age	0.98	0.79, 1.22	0.9
Females [vs males)]	12.5	0.58, 270	0.11
Valve surgeries alone [vs CABG +/− valve]	3.16	0.19, 53.6	0.4
EtCo2	0.98	0.97, 1	0.024
Isoflurane concentration	1.03	0.64, 1.63	>0.9

¹CI = Confidence Interval

^*p-values for categorical variables of more than two categories are shown

Compared to pre-CPB, odds of cerebral desaturation were higher during CPB (22.1 [12.4–39.2], p<0.001) and post-CPB (18.2 [12.2–27.3], p<0.001).

Associations for concurrent cerebral desaturation and burst suppression (Table 4)

Table 4.

Factors associated with intraoperative concurrent cerebral desaturation and burst suppression

Characteristic	OR.	95% CI1	p-value	p-value (Type III)*
CPB [vs pre-CPB]	52.3	19.5, 140	<0.001	< 0.001
Post CPB [vs pre-CPB]	12.7	6.39, 25.2	<0.001
Age	1.16	0.88, 1.52	0.3
Female [vs male]	4.59	0.13, 156	0.4
Valve surgeries [vs CABG +/− Valve]	16.9	0.62, 460	0.094
EtCo2	1	0.98, 1.02	>0.9
Isoflurane concentration	2.81	1.33, 5.92	0.007

¹CI = Confidence Interval

^*p-values for categorical variables of more than two categories are shown

Compared to pre-CPB, the odds of concurrent cerebral desaturation and burst suppression were increased during CPB (52.2 [19.5, 140], p<0.001) and post-CPB (12.7 [6.39–25.2], p<0.001) (figure 1 from R1). (Table 4) An increase (per 0.1%) in isoflurane concentration was associated with higher odds of concurrent cerebral desaturation and burst suppression : 2.81 [1.33–5.92], p=0.007.

Associations for burst suppression during CPB (Supplemental table 3)

During CPB, the odds of burst suppression were increased with cerebral desaturation, (compared to normal cerebral oximetry); OR 1.93, [1.06–3.53], p=0.033, and with higher concentrations (per 0.1% increase) of isoflurane (1.94, [1.28– 2.95], , p<0.001). The period between aortic cross-clamp removal and end of CPB did not show a statistically significant difference in the odds of burst suppression compared to the rest of the CPB period (1.42, [1.92–2.2], p=0.11).

Associations for cerebral desaturation during CPB (Supplemental table 4)

Compared to the rest of the CPB, the odds of cerebral desaturation were higher during the the period between aortic cross-clamp removal and end of CPB (7.79 [4.02–15.1, p<0.001).

Associations for concurrent cerebral desaturation and burst suppression within CPB (Supplemental table 5)

Compared to the rest of the CPB, the odds of concurrent cerebral desaturation and burst suppression were significantly higher during the period between aortic cross-clamp removal and the end of CPB (10 [4.01–25.1], p<0.001). The odds of concurrent abnormalities was not different by gender. An increase (per 0.1%) in isoflurane concentration during CPB was associated with higher odds of concurrent abnormalities 6.2 [2.13–18], p<0.001.

Discussion

This study aimed to explore the intraoperative incidence of and associations between cerebral desaturation and burst suppression. First, we quantified the occurrence of cerebral desaturation, burst suppression, and their concurrency. Second, we report an association between cerebral desaturation with burst suppression. Last, we found that cerebral desaturations as well as, concurrent cerebral desaturations and burst suppression were more likely during CPB, particularly during the the period between removal of aortic cross-clamp and end of CPB.

The association of cerebral desaturations and burst suppression with postoperative cognitive disorders has been studied extensively in the past^20,21,8. Previous studies have proposed combining raw EEG and cerebral oximetry into perioperative brain health monitoring⁹,²². The association between intraoperative cerebral desaturations and burst suppression, and their concurrence, has not been explored yet. Our findings suggest that inhaled anesthetic concentration as well as cerebral desaturation are strongly associated with intraoperative burst suppression. This is in contrast to attributing intraoperative burst suppression solely to an increased depth of anesthesia²³ and emphasizes the need for further exploration of simultaneous monitoring of cerebral oximetry and EEG in cardiac surgeries. Additionally, the strong association between anesthetic concentration and burst suppression underscores the importance of monitoring burst suppression alongside end-tidal concentrations of inhalational anesthetics.

The likelihood of cerebral desaturation and, concurrent cerebral desaturation and burst suppression was notably higher during the CPB period compared to the pre-CPB period. Previous studies have shown increased incidence of embolism in on-pump cardiac surgeries compared to off-pump indicating the contribution of CPB to embolism^13,24. However, cerebral hypo-perfusion that can potentiate postoperative neurocognitive dysfunction can result from factors other than emboli, such as non-pulsatile blood flow, hemodynamic instability, CPB flow settings, rewarming, and increased vasopressor use, which are common during CPB²⁵. Hence, the contribution of CPB to postoperative cognitive dysfunction cannot be fully explained by cerebral embolism. A study of cerebral perfusion using optical methods showed a significant reduction in cerebral blood flow and metabolism during the transition to CPB¹⁴. Our findings of higher occurrence of cerebral desaturation during the CPB period demonstrates the association between CPB and possible cerebral hypo-perfusion. Moreover, the concurrence of cerebral desaturation and burst suppression during CPB indicates that the CPB period is possibly associated with clinically significant cerebral hypo-perfusion that may lead to postoperative cognitive dysfunction.

The phase reflective of the period between aortic cross-clamp removal and the end of CPB showed the highest odds for cerebral desaturations, and concurrent cerebral desaturations and burst suppression. Previous literature showed that periods of aortic cross-clamping and cross-clamp removal were associated with the highest rate of cerebral emboli¹³. However, we did not investigate this. Equally important factors to consider are hemodynamic instability, for example, periods when reductions in CPB flow are requested to an absolute minimum. Further exploration by analyzing the concurrence of these brain health abnormalities with potential causal factors during each phase of the surgery might better establish the etiologies more clearly.

Our study has limitations. The study included patients at one hospital only, and this limits the generalizability of our findings. It is a retrospective study and no causal inferences are possible. The anesthetic depth was not recorded. Cerebral oximetry through NIRS is not a precise approximation of cerebral perfusion, which is multifactorial. A priori sample size calculation was not conducted, and a convenience sample was utilized. Since EEG and cerebral oximetry data collection were blinded, instances of sensor detachment and/or high impedance of the electrodes were not addressed immediately, leading to interrupted data collection and missing data. Other factors potentially affecting cerebral pefusion (and cerebral desaturation) such as MAP, hemoglobin levels, blood transfusions, and perioperative medications were not considered in the analysis. These factors, along with the missing data, introduce the potential for bias. There is an inherent chance of Type I errors given the three primary outcomes, several independent variables of main interest, overlap in data between the entire surgery and CPB phase analyses, and multiple comparisons from testing the same hypotheses across different models,.

Conclusion

In cardiac surgical patients, there was a significant concurrence between cerebral desaturation and burst suppression, particularly during CPB, These findings suggest an interaction between a presumed cerebral hypo-perfusion with intraoperative burst suppression. Future efforts should be focused on elucidating the underlying mechanisms behind burst suppression and developing appropriate interventions to improve postoperative cognition.

Figure 1b.

Average duration of cerebral desaturation, burst suppression, and concurrent events before, during, and after cardiopulmonary bypass. Average durations per patient (minutes)

KEY POINTS SUMMARY.

• Question:

  • Are intraoperative EEG burst suppression and cerebral desaturation associated in cardiac surgical patients, and in which phase of cardiac surgery is this association predominant?
• Findings:

  • In cardiac surgical patients, intraoperative EEG burst suppression was associated with concurrent cerebral desaturation, most significantly during cardiopulmonary bypass (CPB), especially during the period between removal of aortic cross-clamp and end of CPB.
• Meaning:

  • Simultaneous monitoring of cerebral oximetry and EEG during cardiac surgery may help identify factors potentially contributing to burst suppression.