Comparison of Remimazolam and Propofol for Anesthesia Management in Liver Cancer Patients Undergoing Percutaneous Radiofrequency Ablation: A Randomized Controlled Trial

Abstract

Background

Ultrasound-guided percutaneous radiofrequency ablation (PRFA) is an effective treatment for small hepatocellular carcinoma, but anesthesia management in patients with liver cancer and comorbidities remains challenging. This study aimed to compare the anesthetic efficacy and safety of remimazolam combined with remifentanil versus propofol in patients undergoing PRFA.

Methods

Eighty-two participants were randomly assigned to receive either propofol (Group P, n = 41) or remimazolam (Group R, n = 41) for anesthesia, both combined with remifentanil for analgesia. The primary outcome was mean arterial pressure (MAP) across seven perioperative time points. Secondary outcomes included heart rate (HR), Cerebral State Index (CSI), oxygen saturation (SpO₂), respiratory parameters, recovery time, and adverse events. Primary analysis followed the intention-to-treat (ITT) principle using linear mixed-effects models. A per-protocol sensitivity analysis was also performed.

Results

In the ITT analysis (n = 82), linear mixed-effects models showed a significant group × time interaction for MAP (F[6, 77.139] = 11.913, p < 0.001), with higher MAP in Group R at T₂, T₄, and T₅. Significant interactions were also observed for CSI and HR, but not for SpO₂. Group R had higher respiratory rate (p = 0.015), lower apnea incidence (17.1% vs 36.6%, p = 0.038), lower P_ETCO₂ (p < 0.001), less hypotension requiring vasopressors (22.0% vs 46.3%, p = 0.015), and longer time to achieve an MOAA/S score ≤1 (19.61 ± 2.68 vs 15.20 ± 1.58 s, p< 0.001). Recovery time was shorter in Group R (9.65 ± 1.81 vs 15.05 ± 3.58 min, p < 0.001), noting flumazenil was given only to Group R. No differences were found in pain scores or liver function tests.

Conclusion

Remimazolam with remifentanil provided better hemodynamic stability and respiratory safety than propofol in PRFA for liver cancer (Child-Pugh A/B, without severe comorbidities). The shorter recovery time, though partly confounded by flumazenil, supports remimazolam as a promising alternative. Further studies in higher-risk patients are needed.

Clinical Trial Registration

The trial was registered before patient enrolment at http://www.chictr.org.cn (ChiCTR2300078192), and was registered on November 30, 2023.

Introduction

Ultrasound-guided percutaneous radiofrequency ablation (PRFA) has emerged as a safe and effective minimally invasive treatment for small hepatocellular carcinoma, characterized by minimal trauma, short operative duration, and rapid postoperative recovery.1,2 However, a substantial proportion of patients with primary or metastatic liver cancer present with concomitant cirrhosis, impaired liver function, insufficient remnant liver volume, or multiple comorbidities, making anesthetic management particularly challenging.3,4 These patients require anesthesia that ensures intraoperative stability, rapid postoperative emergence, and minimal anesthesia-related complications.

Currently, intravenous anesthesia with propofol or dexmedetomidine combined with opioids such as remifentanil is commonly employed for PRFA,5–7 providing adequate sedation and analgesia while allowing patient cooperation during the procedure. Studies have demonstrated that propofol combined with remifentanil is safe and reliable for anesthesia management in liver PRFA.6,7 Nevertheless, propofol exhibits dose-dependent cardiorespiratory depressant effects.8–10 At higher doses, propofol reduces myocardial contractility, and decreases systemic vascular resistance, which may result in reduced cardiac output, relative hypovolemia, and significant fluctuations in circulatory and respiratory function.11 Liu et al have confirmed that propofol and dexmedetomidine can inhibit sympathetic nerve activity, but midazolam has no effect on sympathetic nerve activity.12 This indicates that benzodiazepine drugs have a certain protective effect on the circulation. Moreover, propofol monotherapy is associated with injection pain and difficulty in controlling sedation depth, particularly in patients with poor baseline status and comorbidities, potentially resulting in hypotension and respiratory depression.13 Consequently, there is an urgent need to explore anesthetic regimens that effectively control stress responses, minimize respiratory depression, maintain stable hemodynamics, and accelerate postoperative recovery.

Remimazolam tosylate is a novel ultrashort-acting benzodiazepine. Structurally derived from midazolam, it incorporates a metabolizable methyl propionate side chain, allowing rapid hydrolysis by tissue esterases into an inactive carboxylic acid metabolite.14 Remimazolam exerts its sedative-hypnotic effects by binding to benzodiazepine receptors, enhancing gamma-aminobutyric acid (GABA) binding to GABA-A receptors, increasing chloride ion influx, hyperpolarizing the cell membrane, and suppressing central nervous system activity.15 Compared with propofol, remimazolam demonstrates minimal impact on hemodynamics and respiratory function and has been successfully applied in procedural sedation as well as induction and maintenance of general anesthesia.9,16,17 Notably, its metabolism is independent of hepatic and renal function, making it particularly suitable for patients with hepatic or renal impairment, elderly individuals, and those with compromised physiological reserves.18 Studies have shown that intravenous administration of remimazolam (0.2–0.4 mg/kg) significantly reduces stress responses after hip replacement or cardiac surgery, with efficacy comparable to propofol.19,20

To date, no studies have investigated the application of remimazolam in patients undergoing PRFA for liver cancer. Therefore, this randomized controlled trial aimed to evaluate the anesthetic efficacy and safety of remimazolam combined with remifentanil compared with propofol combined with remifentanil in patients undergoing PRFA for liver cancer. While our study focuses on a relatively well-defined population (Child-Pugh A/B without severe comorbidities) to establish initial safety and efficacy data, the broader challenge of anesthetic management in high-risk patients with significant comorbidities or Child-Pugh C cirrhosis warrants further investigation.

Methods

Study Design and Ethics

This single-center, randomized, assessor-and patient-blinded, parallel-group controlled trial was conducted at the Affiliated Hospital of North Sichuan Medical College, China, from September 2024 to August 2025. The study protocol was approved by the Medical Ethics Committee of the Affiliated Hospital of North Sichuan Medical College (Approval No. 2023ER3791) and registered with the Chinese Clinical Trial Registry (ChiCTR2300078192). The study was performed in accordance with the principles of the Declaration of Helsinki and adhered to the Consolidated Standards of Reporting Trials (CONSORT) guidelines. Written informed consent was obtained from all participants prior to enrollment.

Participants and Randomization

This study included a total of 82 patients who were scheduled to undergo ultrasound-guided percutaneous radiofrequency ablation. The inclusion criteria were American Society of Anesthesiologists (ASA) physical statusII–III, age 30–70 years, Body mass index (BMI) <30 kg/m². The exclusion criteria as follows:

History of hypertension, coronary heart disease, diabetes mellitus, or central nervous system disorders; severe cardiopulmonary dysfunction; known allergy to propofol, remimazolam, benzodiazepines, or opioids; pregnancy or lactation. If the surgery duration exceeded 2 hours, change in anesthesia or surgical approach during the procedure or patient request for withdrawal, they will be withdrawn from the trial.

Eligible patients were randomly assigned to either the propofol group (Group P) or the remimazolam group (Group R) in a 1:1 ratio using computer-generated random block numbers (block size of 4) without stratification to ensure balanced group allocation. The randomization sequence was concealed in sequentially numbered, opaque, sealed envelopes prepared by an independent research assistant not involved in patient care. The attending anesthesiologist (who was not blinded to group allocation due to the nature of drug preparation) opened the envelope before induction and prepared the study drugs accordingly. Postoperative assessments were performed by an independent researcher blinded to group allocation. Patients were also blinded to the treatment assignment.

Anesthesia Protocol

All patients fasted for 6 hours from solids and 2 hours from clear liquids before surgery. No premedication was administered. Upon arrival in the operating room, ambient temperature was maintained at 24–26°C. An intravenous line was established, and lactated Ringer’s solution was infused at 20 mL/kg initially, followed by a maintenance rate of 15 mL/kg/h. Standard monitoring included non-invasive systolic blood pressure (SBP), diastolic blood pressure (DBP), electrocardiography, heart rate (HR), and pulse oximetry (SpO₂). The Cerebral State Index (CSI) was continuously monitored using three electrodes placed on the right frontal, temporal, and occipital regions. Oxygen was administered via face mask at 3 L/min. In the propofol group, anesthesia was induced with a slow intravenous injection of propofol 2 mg/kg. In the remimazolam group, anesthesia was induced with a slow intravenous injection of remimazolam 0.2 mg/kg. In both groups, remifentanil was administered via target-controlled infusion (TCI) at an effect-site concentration of 1 ng/mL. When the modified Observer’s Assessment of Alertness/Sedation (MOAA/S) score21 decreased to ≤1, a laryngeal mask airway (LMA) was inserted. Intraoperatively, remifentanil was maintained at a target-controlled effect-site concentration of 1–2 ng/mL in both groups. Propofol was continuously infused at a rate of 4–10 mg/kg/h, and remimazolam at 1–2 mg/kg/h. If body movement occurred, a bolus dose of propofol 1 mg/kg or remimazolam 0.1 mg/kg was administered immediately, depending on the group assignment.

The CSI was maintained between 40 and 60 throughout the procedure. If mean arterial pressure (MAP) decreased by more than 30% from baseline or less than 60mmHg, ephedrine (6 mg) or another vasopressor was administered. If apnea (absence of spontaneous breathing >20 seconds) or SpO₂ < 90% occurred, manual-assisted ventilation was provided until spontaneous breathing recovered or SpO₂ normalized. At the end of surgery, the infusion of all intravenous anesthetic drugs were discontinued immediately. Group R received flumazenil 0.2 mg intravenously, while Group P received an equal volume of normal saline. The LMA was removed when the patient resumed the ability to open eyes or nod on command. Then the patient is transferred to the post-anesthesia care unit (PACU).

After the patients entered the PACU, researchers who had received specialized training and were not familiar with grouping situation conducted Visual Analogue Scale (VAS) pain assessments for both groups of patients at the time of entering the PACU, 15 minutes later, and 30 minutes later (0 points: no pain; 1–3 points: slight pain, tolerable; 4–6 points: patients experienced pain that affected sleep but could still tolerate it; 7–10 points: patients had gradually increasing pain, unbearable pain). If the VAS score is 4 or higher, oxycodone at a dose of 0.1mg/kg will be administered for pain relief. Once the patient’s modified Aldrete post-anesthesia score22 is greater than or equal to 9, the patient is returned to the ward.

Outcome Measures

The primary outcomes was MAP, measured at the following seven time points: before anesthesia induction (T₁), 1 minute after anesthesia induction (T₂), during laryngeal mask insertion (T₃), during puncture (T₄), during radiofrequency ablation (T₅), at the end of the surgery (T₆), and upon awakening (T₇).

Secondary outcomes included the following: 1. Heart rate (HR), Cardiac Stress Index (CSI), and peripheral oxygen saturation (SpO₂) at T₁–T₇. 2. Respiratory rate and tidal volume at T₂; the number of patients experiencing apnea longer than 20 seconds and SpO₂ less than 90%; and end-tidal carbon dioxide partial pressure (P_ETCO₂) at T₅. 3. Anesthesia duration (from the induction of anesthesia to the cessation of drug administration), the time to reach MOAA/S ≤1 (the time to reach an MOAA/S score ≤1 from fully awake following intravenous injection of remimazolam or propofol), operation duration (from start of puncture to end of surgery), awakening time (from cessation of anesthetic drugs to removal of laryngeal mask), time of residence in the PACU (the time from entering the PACU to leaving the PACU). 4. Total dosage of propofol, total dosage of remifentanil, total dosage of remimazolam, and usage of vasopressor drugs for treating hypotension; 5. VAS scores in the PACU; postoperative oxycodone dosage required. 6. Incidence of intraoperative movement, intraoperative awareness, intraoperative allergy, postoperative nausea and vomiting (PONV), and surgeon satisfaction (the 10-point scale is used to assess the satisfaction of surgeons with the anesthetic effect, 0–3 points defined as unsatisfactory, 4–7 points as relatively satisfactory, and 8–10 points as satisfactory). 7. Blood samples (2 mL) were collected preoperatively and on postoperative days 2 and 5 to determine serum levels of liver function indicators, such as aspartate aminotransferase (AST) and alanine aminotransferase (ALT).

Sample Size Calculation

The sample size for this study was calculated using GPower 3.1 software. Based on MAP data from a previous study by Guo et al,23 the mean and standard deviation (SD) for the trial group after anesthesia induction were 72.1 and 4.9, respectively, while the mean and SD for the control group were 68.0 and 4.1, respectively. Using these values, the calculated effect size was d = 0.9. Taking into account a 10% dropout rate, the final estimated sample size is 82 participants (41 per group).

Statistical Analysis

Statistical analysis was performed using GraphPad Prism 9.2.0 (GraphPad Software, San Diego, CA, USA) and SPSS 23.0 (IBM Corp., Armonk, NY, USA). Normality of continuous data was assessed using the Shapiro–Wilk test. Continuous variables were expressed as mean ± SD and compared between groups using independent-samples t-test. Categorical variables were presented as frequencies (percentages) and compared using the chi-square test.

For all repeated-measures outcomes (MAP, HR, CSI, SpO₂, ALT and AST), linear mixed-effects models (LMM) were applied to the intention-to-treat (ITT) population (n = 82) according to the original randomization allocation. The models included treatment group, time, and group × time interaction as fixed effects, with patient as a random intercept. This approach utilizes all available repeated measurements and is robust to missing observations under the missing-at-random assumption. Estimated marginal means (EMMs) and their standard errors (SE) derived from the LMMs were used for data description and graphical presentation. Post-hoc pairwise comparisons between groups at each time point were adjusted using the Bonferroni method. A per-protocol (PP) sensitivity analysis was additionally conducted for the primary outcome, excluding the three patients who discontinued intervention due to a change in surgical procedure, to confirm the robustness of the findings. A two-sided p value < 0.05 was considered statistically significant.

Results

A total of 145 patients were assessed for eligibility, of whom 63 were excluded (32 did not meet inclusion criteria, 18 declined to participate, 13 for other reasons). The remaining 82 patients were randomized to Group P (propofol, n=41) or Group R (remimazolam, n=41). After randomization, two patients in Group P and one patient in Group R discontinued intervention due to a change in surgical procedure. In accordance with the intention-to-treat (ITT) principle, all 82 randomized patients were included in the primary analysis. Missing data for the primary outcome were handled using a linear mixed-effects model, which incorporates all available repeated measurements without requiring complete data at every time point. A per-protocol (PP) sensitivity analysis was additionally performed, comprising 39 patients in Group P and 40 patients in Group R who completed the protocol-defined intervention (Figure 1).

Figure 1.

CONSORT 2025 flow diagram.

There were no significant differences between the two groups in terms of age, sex, BMI, ASA status, Child-Pugh grade,24 tumor size, and the number of tumors (p > 0.05) (Table 1).

Table 1.

Baseline Characteristics of Study Participants (ITT Population, n = 82)

Characteristic	Group P (n=41)	Group R (n=41)	t/χ2	Difference (95% CI)	p
Age (years)	56.10 ± 6.79	54.15 ± 6.04	−1.374	−1.95 (−4.78 to 0.88)	0.173
Sex (male/female)	35/6	30/11	1.855	−12.2% (−29.5% to 5.2%)	0.173
BMI (kg/m2)	23.66 ± 1.24	23.51 ± 1.56	−0.471	−0.15 (−0.77 to 0.47)	0.639
ASA status (II/III)	13/28	19/22	1.845	+14.6% (−6.2% to 35.5%)	0.174
Child-Pugh grade (A/B)	16/25	22/19	1.766	+14.6% (−6.7% to 36.0%)	0.184
Tumor size (cm)	4.79 ± 1.43	4.23 ± 1.22	−1.656	−0.56 (−1.15 to 0.02)	0.102
Number of tumors (single/multiple)	28/13	30/11	0.236	+4.9% (−14.8% to 24.5%)	0.627

Notes: Data are presented as mean ± SD or n. For continuous variables, differences are presented as Group R minus Group P with 95% confidence intervals (CIs) calculated from independent-samples t-tests. For categorical variables, differences are presented as risk differences (RD) with 95% CIs. No significant differences were observed between groups.

Abbreviations: BMI, body mass index; ASA, American Society of Anesthesiologists.

Changes in the primary outcome (MAP) and secondary outcomes (HR, SpO₂, and CSI) over time are shown in Figure 2. For the primary outcome, a linear mixed-effects model was fitted to the intention-to-treat population (n = 82), incorporating all available MAP measurements. The analysis revealed a significant group × time interaction for MAP (F[6,77.139] = 11.913, p < 0.001, partial η² = 0.481). Post-hoc pairwise comparisons with Bonferroni adjustment showed that MAP was significantly higher in Group R than in Group P at T₂, T₄, and T₅. A per-protocol (PP) sensitivity analysis (n = 79) yielded consistent results with the ITT analysis (F[6,77] = 12.225, p < 0.001, partial η² = 0.488), confirming the robustness of the primary findings.

Figure 2.

Changes in hemodynamic and respiratory parameters over time. (A) Mean arterial pressure (MAP, mmHg); (B) Heart rate (HR, beats per minute); (C) Cerebral State Index (CSI); (D) Peripheral oxygen saturation (SpO₂, %). Data are presented as estimated marginal means ± standard error (SE) from the intention-to-treat (ITT) analysis using SPSS EM Means. Time points: T₁, before induction; T₂, 1 min after induction; T₃, at laryngeal mask airway insertion; T₄, at puncture; T₅, during radiofrequency ablation; T₆, at the end of surgery; T₇, upon awakening. Group R (remimazolam) is shown with black circles and solid lines; Group P (propofol) with pink squares and dashed lines. *Indicates a significant difference between Group R and Group P at the same time point (p < 0.05, adjusted for multiple comparisons using Bonferroni correction).

For secondary outcomes, linear mixed-effects models were applied to the intention-to-treat population (n = 82). A significant group × time interaction was observed for CSI (F[6,31.520] = 14.828, p < 0.001, partial η² = 0.738) and HR (F[6,79.605] = 3.641, p =0.003, partial η² = 0.215) but not for SpO₂ (F[6,86.844] = 1.815, p =0.105, partial η² = 0.109). CSI values were significantly lower in Group P compared with Group R at T₂ through T₆ (p < 0.05). At T₂, HR was significantly lower in Group R than in Group P (p < 0.05). No significant differences in SpO₂ were observed between groups at any time point.

Respiratory parameters, postoperative pain, and adverse events are summarized in Table 2. At T₂, the respiratory rate was significantly higher in Group R than in Group P (p = 0.015). Tidal volume did not differ significantly between groups (p = 0.362). The incidence of apnea (>20 seconds) was significantly lower in Group R compared with Group P (p = 0.038). P_ETCO₂ during T₅ was significantly lower in Group R (p < 0.001). Postoperative VAS scores at PACU admission, 15 min, and 30 min showed no significant differences between the two groups (p > 0.05). The proportion of patients requiring rescue oxycodone (VAS ≥4) was similar between groups (p = 0.435). The incidence of hypotension requiring vasopressor treatment was significantly lower in Group R than in Group P (p = 0.015). The incidences of intraoperative body movements and PONV were similar between the two groups (p > 0.05).

Table 2.

Respiratory Parameters, Postoperative Pain, and Adverse Events (ITT Population, n = 82)

Parameter		Group P (n=41)	Group R (n=41)	t/χ2	Difference (95% CI)	p value
Respiratory rate (breaths/min) at T2		12.39 ± 3.18	14.37 ± 3.97	2.487	1.98 (0.39 to 3.56)	0.015
Tidal volume (mL) at T2		289.10 ± 74.99	304.07 ± 68.78	0.349	14.98 (−16.65 to 46.60)	0.362
Apnea >20s (n, %)		15 (36.6%)	7 (17.1%)	4.318	−19.5% (−38.2% to −0.8%)	0.038
PETCO2 at T5 (mmHg)		52.03 ± 5.82	46.28 ± 7.45	−3.816	−5.75 (−8.75 to −2.75)	<0.001
VAS score	PACU admission	3.28 ± 0.94	3.23 ± 0.86	−0.281	−0.05 (−0.46 to 0.35)	0.780
	15 min	2.74 ± 0.79	2.45 ± 0.60	−1.874	−0.29 (−0.61 to 0.02)	0.065
	30 min	1.90 ± 0.68	1.70 ± 0.65	−1.321	−0.20 (−0.50 to 0.10)	0.191
Patients requiring oxycodone (n, %)		19 (46.3%)	16 (39.0%)	0.608	−7.3% (−28.7% to 14.0%)	0.435
Hypotension requiring vasopressor (n, %)		19 (46.3%)	9 (22.0%)	5.932	−24.4% (−44.2% to −4.6%)	0.015
Intraoperative body movements (n, %)		2 (4.9%)	4 (9.8%)	0.668	+4.9% (−6.3% to 16.1%)	0.414
PONV (n, %)		6 (14.6%)	8 (19.5%)	0.288	+4.9% (−11.4% to 21.1%)	0.591
Allergy (n, %)		0	2 (4.9%)	2.001	+4.9% (−1.7% to 11.5%)	0.157

Notes: Data are presented as mean ± SD or n (%). For continuous variables, differences are presented as Group R minus Group P with 95% confidence intervals (CIs) calculated from independent-samples t-tests. For categorical variables, differences are presented as risk differences (RD) with 95% CIs. For adverse events (apnea, hypotension, body movements, PONV, and allergy), the three patients who discontinued intervention were conservatively assumed to have no event for the intention-to-treat analysis (n = 82). For P_ETCO₂ and VAS scores, data were available for 79 patients (39 in Group P and 40 in Group R) due to missed postoperative assessments in the three patients who discontinued intervention. T₂, 1 min after induction; T₅, during ablation.

Abbreviations: P_ETCO₂, end-tidal carbon dioxide; VAS, Visual Analog Scale; PACU, post-anesthesia care unit; PONV, postoperative nausea and vomiting.

As shown in Table 3, recovery time was significantly shorter in Group R than in Group P (p < 0.001). The time to achieve an MOAA/S score ≤1 was significantly longer in Group R than in Group P (p < 0.001). Total remifentanil dose and physician satisfaction scores did not differ between groups.

Table 3.

Intraoperative and Recovery Characteristics (ITT Population, n = 82)

Parameter	Group P (n=41)	Group R (n=41)	t	Difference (95% CI)	p value
Anesthesia duration (min)	50.41 ± 11.44	48.75 ± 9.39	−0.706	−1.66 (−6.34 to 3.02)	0.482
Operation duration (min)	25.41 ± 9.59	24.68 ± 7.74	−0.376	−0.74 (−4.63 to 3.16)	0.708
Recovery time (min)	15.05 ± 3.58	9.65 ± 1.81	−8.442	−5.40 (−6.68 to −4.12)	<0.001
The time to reach MOAA/S ≤1 (s)	15.20 ± 1.58	19.61 ± 2.68	9.074	4.41 (3.45 to 5.38)	<0.001
Time of residence in the PACU	37.26 ± 3.90	36.93 ± 3.66	−0.39	−0.33 (−2.02 to 1.36)	0.698
Total propofol dose (mg)	296.64 ± 79.63	N/A
Total remimazolam dose (mg)	N/A	74.50 ± 10.57
Total remifentanil dose (μg)	359.49 ± 47.46	349.75 ± 54.76	−0.844	−9.74 (−32.72 to 13.24)	0.401
Physician’s satisfaction	8.44 ± 0.94	8.35 ± 1.12	−0.368	−0.09 (−0.55 to 0.38)	0.714

Notes: Data are presented as mean ± SD. Differences are presented as Group R minus Group P with 95% confidence intervals (CIs). MOAA/S, Modified Observer’s Assessment of Alertness/Sedation. For anesthesia duration, operation duration, recovery time, time of residence in the PACU, total remifentanil dose and physician’s satisfaction, data were available for 79 patients (39 in Group P and 40 in Group R) due to missed postoperative assessments in the three patients who discontinued intervention.

Table 4 summarizes the perioperative changes in liver function indicators (AST and ALT) for both groups. For the intention-to-treat population (n = 82), linear mixed-effects models revealed no significant group × time interaction for AST or ALT. Similarly, no significant main effect of Group was observed.

Table 4.

Liver Function Indicators Were Measured at 2d and 5d After the Operation (ITT Population, n = 82)

Indicators		Preoperative	2d Postoperative	5d Postoperative
AST (U/L)	Group P (n = 41)	49.88 ± 1.95	375.65 ± 15.76	57.92 ± 1.00
	Group R (n = 41)	48.54 ± 1.94	390.16 ± 15.56	55.78 ± 0.99
	Group: F (1, 93.277) = 0.249, p = 0.619, partial η2 = 0.003
	Time: F (2, 95.051) = 533.451, p < 0.001, partial η2 = 0.918
	Group×Time: F (2, 95.051) = 0.267, p = 0.766, partial η2 = 0.006
ALT (U/L)	Group P (n = 41)	55.32 ± 3.61	446.90 ± 11.70	152.74 ± 7.05
	Group R (n = 41)	58.34 ± 3.61	461.60 ± 11.55	151.33 ± 6.96
	Group: F (1, 146.635) = 0.662, p = 0.417, partial η2 = 0.004
	Time: F (2, 174.014) = 1088.311, p < 0.001, partial η2 = 0.926
	Group×Time: F (2, 174.014) = 0.365, p = 0.695, partial η2 = 0.004

Notes: Values are estimated marginal means ± SE derived from linear mixed-effects models.

Abbreviations: AST, aspartate aminotransferase; ALT, alanine aminotransferase.

No intraoperative awareness or severe allergic reactions were observed in either group. However, mild allergic reactions (skin rash) were observed in two patients in the remimazolam group (Group R) and none in the propofol group. These symptoms improved after the administration of a small dose of dexamethasone.

Discussion

In this randomized, assessor- and patient-blinded controlled trial, we compared the efficacy and safety of remimazolam tosylate versus propofol, both in combination with remifentanil, in patients undergoing ultrasound-guided PRFA for liver cancer. The principal findings of this study are that remimazolam-based anesthesia provided more stable hemodynamics, reduced respiratory depression, and facilitated faster postoperative recovery compared with propofol, while maintaining comparable analgesic efficacy and overall procedural satisfaction.

Hemodynamic stability is a critical concern in patients undergoing PRFA, particularly in those with underlying liver dysfunction and limited physiological reserve. In the present study, MAP decreased significantly at multiple intraoperative time points in the propofol group, whereas it remained relatively stable in the remimazolam group. Moreover, the incidence of hypotension requiring vasopressor support was significantly lower with remimazolam. These findings are consistent with the pharmacological profile of remimazolam, which, as a benzodiazepine, exerts minimal inhibitory effects on myocardial contractility and systemic vascular resistance.25 Propofol is known to suppress sympathetic tone and induce vasodilation, leading to dose-dependent hypotension,26 which may be particularly detrimental in patients with cirrhosis or compromised cardiovascular function.27,28 Both Tang et al29 and Guo et al23 studies have shown that remimazolam can provide more stable hemodynamic conditions compared to propofol. Hasegawa et al30 discovered through a randomized controlled trial that compared with propofol, remimazolam can maintain the balance of sympathetic and parasympathetic nerves and keep hemodynamic stability. Taken together, these results indicate that remimazolam may offer a safer hemodynamic profile for high-risk patients undergoing minimally invasive liver procedures.

Tang et al31 found that there was no difference in the sedation scores between remimazolam and propofol in critically ill patients. In this study, CSI values were lower in the propofol group, suggesting deeper levels of cortical suppression. Yang et al32 found that the BIS level maintained by remimazolam in patients under general anesthesia was higher than that maintained by propofol, which is consistent with the results of this study. However, the remimazolam group demonstrated a longer time to reach MOAA/S ≤1, indicating a slower onset of deep sedation compared with propofol. Clinically, this slower onset may provide a smoother transition into deep sedation, potentially reducing abrupt hemodynamic fluctuations. Despite this difference, both anesthetic regimens achieved adequate sedation for successful laryngeal mask insertion and procedural completion. Indicating that remimazolam provides sufficient sedative depth when appropriately titrated.

Respiratory safety is a major concern during sedation, especially in procedures requiring spontaneous ventilation.33,34 In this study, remimazolam demonstrated a clear advantage in respiratory safety. The respiratory rate at T₂ was higher in the remimazolam group, and the incidence of apnea episodes was significantly lower compared with the propofol group. Additionally, lower P_ETCO₂ levels during ablation suggest more effective ventilation. These findings align with previous reports that benzodiazepines, including remimazolam, have a more favorable respiratory profile compared with propofol.33

Rapid recovery is essential in minimally invasive procedures such as PRFA to facilitate early postoperative assessment and reduce PACU stay. In this study, the remimazolam group exhibited significantly shorter awakening times compared with the propofol group. This rapid recovery can be attributed to the unique metabolism of remimazolam, which is rapidly hydrolyzed by tissue esterases into inactive metabolites, independent of hepatic or renal function.35 Stöhr T et al18 believe that for patients with liver and kidney dysfunction, intravenous injection of remimazolam will not cause any related side effects. It is important to note, however, that recovery time in the remimazolam group was confounded by the administration of flumazenil (0.2 mg) at the end of surgery, which actively reverses benzodiazepine sedation. This practice, while clinically relevant in our protocol, is not universal across all centers and represents a major confound in interpreting the recovery advantage. Without flumazenil reversal, the recovery time difference may have been smaller, as suggested by remimazolam’s pharmacokinetic profile (terminal half-life of approximately 70–90 minutes).35 The research results of Zhang et al36 confirmed that the combined use of flumazenil and remimazolam for total intravenous anesthesia led to rapid and reliable recovery of consciousness. This characteristic is particularly advantageous in patients with liver dysfunction, where drug accumulation may delay recovery when using traditional anesthetics. Therefore, our finding of significantly shorter recovery time should be interpreted as reflecting the combined effect of remimazolam plus active reversal rather than a property of remimazolam alone. Future studies should evaluate recovery profiles without routine flumazenil administration to better characterize remimazolam’s intrinsic recovery characteristics. Interestingly, despite the shorter recovery time in Group R, PACU residence time was nearly identical between groups. This may be because the modified Aldrete score assesses multiple domains (consciousness, activity, circulation, respiration, and oxygen saturation). The additional time required for consciousness recovery in Group P may not have been the rate-limiting factor for overall Aldrete score achievement.

No significant differences were observed between the two groups regarding serum AST and ALT levels across the three perioperative time points (preoperatively, and on days 2 and 5 postoperatively). Although both groups exhibited a marked transient elevation in transaminases at 2 days postoperatively—reflecting acute hepatic stress induced by the surgical trauma itself—remimazolam did not exacerbate this injury, and the recovery pattern by day 5 was highly consistent with that of the propofol group. This suggests that, despite remimazolam being hydrolyzed by plasma esterases and partially dependent on hepatic metabolism, it does not exert a greater impact on hepatocellular membrane integrity than propofol, indicating a favorable hepatic safety profile. The results of this study are highly consistent with those of Xing et al,37 but further confirmation is still needed through a large number of preclinical studies.

Furthermore, no significant between-group differences were found in postoperative pain scores or the requirements for rescue analgesics, suggesting that perioperative opioid consumption was well-balanced between the two groups. This balance effectively ruled out the confounding effects of inadequate analgesia or differential opioid use on the incidence of PONV. Lastly, surgeon satisfaction scores regarding intraoperative sedation depth and operating conditions did not differ significantly between the two groups, indicating that remimazolam provides equally satisfactory muscle relaxation and hemodynamic stability as propofol, thereby meeting the clinical requirements for surgical procedures.

Although two cases of mild allergic reactions (skin rash) were observed in the remimazolam group, these were transient and resolved with corticosteroid administration, indicating an overall acceptable safety profile.

Several limitations should be acknowledged. First, this was a single-center study with a relatively small sample size, which may limit generalizability. Second, the anesthesiologist was not blinded due to the nature of drug preparation, which may introduce performance bias. Third, the use of flumazenil exclusively in the remimazolam group may have influenced recovery outcomes, potentially exaggerating the difference in awakening time. Finally, long-term outcomes and cognitive recovery were not assessed.

Conclusions

In patients undergoing PRFA for liver cancer with Child-Pugh A/B status and without severe comorbidities, remimazolam combined with remifentanil provided superior hemodynamic stability and improved respiratory safety compared with propofol-based anesthesia. Recovery time was also significantly shorter in the remimazolam group; however, this finding should be interpreted with caution, as flumazenil was administered only to this group. Although the onset of deep sedation was slightly slower with remimazolam, it remained clinically acceptable and may have contributed to smoother intraoperative hemodynamic control. These findings support remimazolam as a promising anesthetic alternative in this setting. Nevertheless, further studies are warranted to confirm its safety and efficacy in higher-risk populations, including those with Child-Pugh C cirrhosis or significant cardiovascular comorbidities.

Funding Statement

This research was supported by the 2025 Annual Scientific Research Development Project of the Clinical Medical College Affiliated Hospital of North Sichuan Medical University (2025MSRJ002) and the “15th Five-Year” Social Science Research Project of Nanchong City for the Year 2026 (NC26B326).

Abbreviations

ITT, intention-to-treat; CONSORT, Consolidated Standards of Reporting Trials; ASA, American Society of Anesthesiologists; BMI, Body mass index; CSI, Cerebral State Index; SBP, systolic blood pressure; DBP, diastolic blood pressure; P_ETCO₂, End-tidal carbon dioxide; GABA, Gamma-aminobutyric acid; HR, Heart rate; MAP, Mean arterial pressure; MOAA/S, Modified Observer’s Assessment of Alertness/Sedation; PACU, Post-anesthesia care unit; PRFA, Percutaneous radiofrequency ablation; PONV, Postoperative nausea and vomiting; SpO₂, Peripheral oxygen saturation; TCI, Target-controlled infusion; VAS, Visual Analog Scale; LMA, laryngeal mask; AST, aspartate aminotransferase; ALT, alanine aminotransferase; LMM, linear mixed-effects models; EMMs, Estimated marginal means; SE, standard errors; SD, standard deviation; PP, per-protocol.

Data Sharing Statement

All data generated or analyzed during this study have been included in the published article. Further inquiries regarding the datasets can be directed to the corresponding author upon reasonable request.

Ethics Approval and Informed Consent

The study was approved by the Institutional Ethics Committee of the Affiliated Hospital of North Sichuan Medical College, China (Approval no. 2023ER3791), and was registered in the Chinese Clinical Trials.gov (No. ChiCTR2300078192). All participants provided written informed consent prior to enrollment. This study complied with the Declaration of Helsinki.

Disclosure

The authors report no competing interests in this work.