EEG Correlates of Delayed Emergence after Remimazolam-Induced Anaesthesia Compared to Propofol

Patient population

This prospective, randomized, double-blinded study was performed between June 2023 and October 2023. The study protocol was approved by the institutional review board of Severance Hospital, Yonsei University Health System. After obtaining ethics committee approval, all patients were informed written consents on the day before surgery and the protocol was done in accordance with relevant guidelines.

The inclusion criteria were as follows: Patients aged 19 to 65 years who meet the American Society of Anaesthesiologists (ASA) physical status classification 1-3 and undergo laparoscopic cholecystectomy. The exclusion criteria were as follows: Pregnant women, individuals with arrhythmias, emergency surgeries, obesity (BMI > 30), day surgery, foreigners, and illiterate individuals.

Study procedure

Among the 60 patients enrolled in this study, individuals with low PSI scores at baseline, insufficient EEG data, or a lack of marker flag were excluded. Consequently, we analysed the data from the remaining 48 patients (n = 24 per group) assigned patients to the remimazolam or propofol group at a 1:1 ratio (Figure 1A). Patients, operators, and two research personnel were unaware of the group identity, but the attending anaesthesiologist couldn’t be blinded about the group identity due to the distinct properties and color of the two anaesthetics.

• Download figure
• Open in new tab

Figure 1.

Schematic diagram. (A) Consort diagram describes the progress of patient via each stage of the randomized trial. (B) Analysis methods describe the progress of the analysis methods which include power spectrogram analysis and functional connectivity measure analysis.

Before entering the operating rooms, we placed the SEDLine sensor on the patient’s forehead. Preoperative baseline EEG were recorded at controlled eye-open (focused wakefulness) and eye-closed (relaxed wakefulness) conditions, for each 4-minute period.

Upon entering the operating rooms, the patients were monitored with pulse oximetry, non-invasive blood pressure, electrocardiography, and anaesthetic depth (SedLine^®, Masimo Corp., Irvine, CA, USA). The patients received 0.1 mg glycopyrrolate before infusing remifentanil and remimazolam or propofol using a commercial syringe pump (Agillia, SB Medica SRL, Italy).

In the propofol group, anaesthesia was induced using propofol [target-controlled infusion (TCI), Marsh model] and remifentanil at effect-site concentrations of 4 mcg ml^-1 and 3 ng ml^-1 respectively. The concentration of propofol is adjusted to maintain an appropriate depth of anaesthesia based on the EEG (PSI target 40) until the end of the surgery. In the remimazolam group, anaesthesia was induced using remifentanil at an effect-site concentration of 3 ng ml^-1 (TCI, Minto model) and remimazolam at a flow rate of 6 mg kg^-1 h^-1, as per the manufacturer’s recommendations. The concentration of remimazolam is adjusted to maintain an appropriate depth of anaesthesia based on the EEG (PSI target 40) until the end of the surgery. More details are provided in Supplementary Method (‘Study Procedure’).

EEG Data pre-processing

A four-channel (Fp1(L1), Fp2(R1), F7(L2), F8(R2)) EEG signals (n = 60; 30 in each group) were acquired during the entire anaesthetic process, including baseline, unconscious, and emergence periods. As EEG characteristics are distinctly exhibited in different anaesthetic states, the data were divided into 6 states. EEGLAB open-source toolbox for MATLAB¹⁹ was used for data extraction by states. (1) baseline eyes-closed (EC): defined as 3 minutes after eyes-closed; (2) baseline eyes-open (EO): defined as 3 minutes after eyes-open; (3) after loss of consciousness (after LOC): defined as 15 minutes after the loss of verbal command response following the administration of propofol or remimazolam; (4) before recovery of consciousness (before ROC): defined as 15 minutes before the extubation point; (5) eyes-closed (EC): defined as 3 minutes after patients closed their eyes at PACU; (6) eyes-open (EO): defined as 3 minutes after patients opened their eyes at the PACU. To compare the prefrontal EEG characteristics of two different anaesthetic agents under the unconsciousness period in stages, the periods after LOC and before ROC were divided into five intervals, aligning with the duration of baseline EO and EC states for significance assessments.

The EEG data of each state was segmented into a 10-sec time series (with a uniform epoch length), which generated a total of 18 epochs (3*60)/10 for each subject and each band.

Prefrontal EEG data analyses

The prefrontal EEG data was analysed using MATLAB and band-pass filtered within the range of 1-50Hz. The analysis included the power spectrogram which was focused on five frequency bands (delta: 0.5-4 Hz, theta: 4-8 Hz, alpha: 8-13 Hz, beta: 13-20 Hz, gamma: 20-40 Hz). Details of power spectrogram analysis and timeseries power analysis are provided in the Supplementary Method (‘Power Spectrogram and Line Power Analyses’). The second section provided in the Supplementary Method (‘Functional Connectivity Measure Analysis’) includes the functional connecitivity analyses (Phase lag entropy and phase lag index) and evaluates the association between the functional connectivity and the clinical scores from propofol- and remimazolam-induced anaesthesia.

Statistical analysis

To evaluate the group differences between propofol-and remimazolam-induced anaesthesia, statistical analyses were performed in RStudio (R version 4.3.2). A two-way repeated-measures analysis of variance was conducted to compare the PSI, power, and functional connectivity between the two groups at 14 different states. Post-hoc analysis with Bonferroni correction was applied to adjust for multiple comparisons. For all data analyses, the corrected P-values of less than 0.05 were deemed to represent a statistically significant difference between the two groups (†P < 0.0001; ***P < 0.001; **P < 0.01; *P < 0.05). Pearson correlation coefficients (r) were calculated using the MATLAB (version R2022b, Mathworks Inc., Co., Ltd., USA) to see the correlation between EEG measures (i.e., connectivity and powers at four frequency bands) and indices values (i.e., PSI scores at PACU (averaged score of EC and EO emergence) and time to Aldrete 9). Benjamini-Hochberg (FDR) correction for multiple testing was applied for the correlations of all comparisons.

Study procedure, participant overview and analytical methodology

Our analysis methods included power spectrogram analysis across various frequency bands, such as delta (0.5-4 Hz), theta (4-8 Hz), alpha (8-13 Hz), beta (13-20 Hz), and gamma (20-40 Hz). Functional connectivity measures were incorporated into the analysis method. We utilized phase lag entropy (PLE) to assess the entropy of relative phase patterns and the phase lag index (PLI) to evaluate phase lead-lag relationships (Figure 1B).

The demographic data of the 24 participants in each group showed no significant differences between the groups, and similar results were observed for the use of remifentanil, operative time, anaesthetic time, blood pressure 5 minutes after intubation, heart rate, and blood pressure 5 minutes before extubation (Table 1). However, patients administered remimazolam exhibited evident sedation and reduced consciousness level at the PACU, reflected by lower Aldrete scores and a RASS, and required a longer duration to meet PACU discharge criteria compared to those receiving propofol (Table 1).

Power spectrogram analysis

In our power spectrogram analysis, remimazolam-induced anaesthesia (Figure 2B) exhibited residual effects (alpha power from 10Hz to 13 Hz) compared to propofol-induced anaesthesia (Figure 2A) during the EO and EC periods of emergence. This power differential was further confirmed in the power (dB) ratio between the two anaesthetic agents (Figure 2C). Around alpha band (8-13Hz), higher propofol power was evident before regaining consciousness (ROC), and higher remimazolam power was noted during emergence (EC, EO) period. Within the beta band (13-20Hz), higher power was consistently observed throughout the unconscious and emergence periods in the remimazolam group. In the gamma band (20-40 Hz), higher power was observed during ROC4-ROC5, as well as during EC and EO emergence in the propofol group. Time-series plots of powers at lower frequency bands exhibited significant differences in anaesthetic drug effects across the anaesthesia process. Specifically, the delta band (Figure 2D) indicated significantly higher power in the propofol group relative to the remimazolam group in the ROC1 (P = 0.039). Theta power (Figure 2E) of propofol was significantly higher throughout the before-ROC period (P = 0.006, P = 0.013, P = 0.004, P < 0.001, P = 0.002), with the opposite observed during the EO emergence period (P = 0.018). Alpha power (Figure 2F) revealed significantly lower levels in the propofol group compared to the remimazolam group during the EC and EO emergence periods (Ps < 0.001). However, no anaesthetic drug difference was observed at the beta and gamma power (Figure 2G, 2H).

• Download figure
• Open in new tab

Figure 2.

Power comparison between propofol (n=24) and remimazolam (n=24) on the prefrontal region (F8). Averaged power spectrograms of (A) propofol-induced anaesthesia and (B) remimazolam-induced anaesthesia at 14 states. (C) Power ratio of two anaesthetic drugs. Time-series plots of powers across the different anaesthesia periods at (D) delta, (E) theta, (F) alpha, (G) beta, and (H) gamma frequency bands. Group and time interaction showed significant effects under all five frequency band powers. †P < 0.0001, ***P < 0.001, **P < 0.01, *P < 0.05 in post-hoc analysis.

Functional connectivity analysis

The results of functional connectivity analyses elucidated distinct prefrontal EEG features between the remimazolam and propofol groups across different frequency bands (Figure 3). PLE, the diversity of phase lead/lag patterns, progressively increased from baseline through the unconscious period in the theta band (4-8 Hz) but decreased in the alpha and beta bands (8-20 Hz). During the unconscious period, the drug difference in PLE was notable across frequency bands, with remimazolam anaesthesia exhibiting higher PLE in the theta and alpha bands, but the opposite finding in the beta band. Specifically, differences were found throughout the after-LOC period (P = 0.006, P = 0.02, P < 0.001, P < 0.001, P < 0.0001) and ROC1-ROC4 (P < 0.001, P = 0.008, P = 0.008, P = 0.036) from the unconscious period in the theta band (Figure 3A); ROC1-ROC4 (P = 0.007, P < 0.001, P < 0.001, P = 0.001) from the unconscious period in the alpha band (Figure 3C); LOC1 (P = 0.015), LOC3-LOC5 (P = 0.008, P = 0.001, P < 0.0001) and ROC1-ROC3 (P = 0.001, P < 0.001, P = 0.017) from the unconscious period in the beta band (Figure 3E). We identified residual anteriorization and delayed arousal during emergence following remimazolam-induced anaesthesia compared to propofol anaesthesia, leading to decreased entropy of phase lead/lag patterns (alpha: P < 0.0001, P = 0.015; beta: P = 0.016, P < 0.001; Figure 3C & 3E).

• Download figure
• Open in new tab

Figure 3.

Phase lag entropy (PLE) and phase lag index (PLI) from prefrontal channel pairs. The measures are averaged over all pairs of 4 channels (Fp1, F8, Fp2, and F8) and subjects. Group comparison throughout the baseline to the emergence periods in PLE and PLI at theta (A, B), alpha (C, D), and beta bands (E, F). The error bars represent the standard error for each state. Blue-filled dots represent the individual values from the propofol-induced group (n=24) and red-filled dots represent the individual values from the remimazolam-induced group (n=24). †P < 0.0001, ***P < 0.001, **P < 0.01, *P < 0.05 in post-hoc analysis.

Conversely, the PLI did not reveal significant differences between the propofol and remimazolam groups in every frequency band upon emergence. During the unconscious period, PLI at the theta frequency band for propofol anaesthesia showed higher values relative to remimazolam anaesthesia in ROC1 (P = 0.044), ROC3 (P = 0.008), and ROC4 (P = 0.007), and PLI at the alpha frequency band showed relatively higher values during the unconscious period compared to the baseline and emergence periods (Figure 3D). Specifically, the propofol group exhibited higher PLI relative to remimazolam anaesthesia across the before-ROC period (P = 0.002, P < 0.001, P < 0.0001, P < 0.0001, P = 0.006). No group difference was observed at the beta band (Figure 3F). Additional results of connectivity measures at the delta and gamma bands are provided in Supplementary Figure S1.

Patient state index and correlation among results

The patient state index (PSI) at different time points exhibited a significant group difference during unconscious and emergence periods. Specifically, the propofol group displayed lower PSI values than those of remimazolam group in ROC2 (P = 0.008) and ROC3 (P = 0.002). Conversely, the PSI values of propofol group were higher than those of remimazolam group during the emergence EC (P < 0.0001) and EO (P = 0.006) states (Figure 4A). Compared with the baseline, PSI values significantly decreased at all time points after anaesthesia intubation in both anaesthetic agents, and the changes in PSI were greater in the propofol group throughout the unconscious and emergence periods.

• Download figure
• Open in new tab

Figure 4.

Patient state index (PSI) and comparisons between post-operative recovery indices and EEG measures. (A) Changes in PSI values across baseline, unconscious, and emergence periods in the propofol and remimazolam groups. The ‘after loss of consciousness’ and ‘before recovery of consciousness’ periods are filled with transparent grey. Shaded regions depict standard deviations. †P < 0.0001, **P < 0.01 in post-hoc analysis.

In Pearson’s correlation results, only the results with significant correlation coefficients after Benjamini-Hochberg (FDR) correction (P < 0.05) were included in Figure 4B-4E and Table 2, with all significance exclusively observed in remimazolam-induced anaesthesia group. The correlation between recovery indices at PACU (measured using the SedLine system) and EEG measures (functional connectivity and power) throughout the anaesthetic stages revealed both positive and negative correlations. PSI at the PACU showed a negative correlation with theta power (4-8 Hz) during the unconscious period, especially in the LOC2- ROC5 states (Figure 4B and Table 2). Conversely, PSI at the PACU demonstrated a positive correlation with alpha band PLE values during the unconscious and emergence periods, particularly at the LOC1, LOC2, ROC1, ROC3, ROC4, EC, and EO states (Figure 4C and Table 2). However, PLE in theta, beta, PLI in theta, alpha, and beta did not show this correlation.

Comparison between time to Aldrete 9 and two functional connectivity measures exhibited opposite results. A strong negative correlation with PLE alpha was observed in LOC5-ROC4 states during the unconscious period (Figure 4D and Table 2). In contrast, a positive correlation with PLI alpha was observed in the ROC1-ROC4 states during the before-ROC period (Figure 4E and Table 2). Pearson’s correlation results show consistency between the functional connectivity findings in the alpha frequency band and sedation of remimazolam anaesthesia at the PACU (Table 1). Various approaches of PSI and time to Aldrete 9 were observed in correlation analysis (all comparisons, Supplementary Tables S1, S2) and are depicted in scatter plots (Supplementary Figures S2, S3). Analgesia Nociception Index (ANI) scores for each anaesthetic agent and correlations between other result values are provided in Supplementary Figure S4 and Table S3. To demonstrate the prediction of delayed emergence, we utilized both the least squares linear regression and random forest regressor models to predict the time to Aldrete 9. The results of prediction models and the detailed explanations are provided in Supplementary Figure S5.

The scatter plots include significant Pearson’s correlations that survived Benjamini-Hochberg (FDR) correction, such as comparison between (B) PSI values at PACU and theta power, (C) PSI values at PACU and phase lag entropy (alpha), (D) time to Aldrete 9 and phase lag entropy (alpha), (E) time to Aldrete 9 and phase lag entropy (alpha). Excluding subjects with a time to Aldrete 9 value of 0, scatter plots of D & E comprise 7 individuals from the propofol group and 17 from the remimazolam group. The edge highlights colored with 3-steps correspond to the significance level of the FDR-corrected P-value (yellow: P < 0.05, orange: P < 0.01, dark orange: P < 0.0001).