This study was approved on 30 January 2022 by The Ethics Committee of Jinhua Hospital Affiliated to Zhejiang University, Zhejiang, China (Protocol Number: 2022–004–001), and the trial was registered at the Chinese Clinical Trial Registry (http://www.chictr.org.cn, ChiCTR2200056891) on 22/02/2022. After receiving approval for clinical registration, the trial was performed from June 2022 to July 2023.

After obtaining informed consent, we enrolled 68 consecutive patients scheduled for LH under LCVP for a minimum of 2 hours at Jinhua Hospital Affiliated to Zhejiang University. The study followed a prospective, randomized, double-blinded design. The inclusion criteria were as follows: patients aged between 18 and 70 years, with American Society of Anesthesiologists (ASA) grade I–II, operated by a single surgeon. The exclusion criteria were as follows: patients with a body mass index (BMI) of more than 35 kg/m2; a history of cranial cerebral trauma surgery, stroke, central nervous system diseases, or psychiatric diseases; primary hypertension > 180 mmHg, uncontrolled hypertension, or postural hypotension; severe heart diseases, such as hypertrophic cardiomyopathy, myocardial infarction, NYHA cardiac function grades III–IV, and echocardiographic of LVEF < 50%. Furthermore, the following patients were also excluded: patients with hepatic decompensation, patients with moderate and severe valve regurgitation and preoperative arrhythmia, dyspnea, or respiratory failure; recent (within 2 weeks) use of heparin, antithrombin, warfarin anticoagulant factor, aspirin antiplatelet aggregation, or other anticoagulant drugs; and known allergy to any drug used in the study.

In accordance with the approved study protocol by the Institutional Review Board, patients were randomly assigned to two groups in a 1:1 ratio using a computer-generated randomization table: the milrinone group and the nitroglycerin group, each comprising 34 patients. The randomization sequences were enclosed within sealed opaque envelopes and managed by two nurse anesthetists who were not involved in the eligibility assessment and recruitment of patients. The nurse anesthetists opened the envelopes and administered the experimental drug once the patients were in the operating room. Throughout the trial, patients, surgeons, and the main researcher responsible for patient anesthesia and follow-up survey data collection remained blinded to the group assignments.

All patients underwent a standardized anesthetic protocol without premedication. In the preoperative preparation room, a peripheral intravenous line was used for all patients by the nurse. Subsequently, internal jugular double-lumen catheterization was performed by the anesthetist in the preparation room under local anesthesia. Fluid administration was restricted to an infusion rate of 2–3 mL/kg/h before anesthesia induction. Upon entering the operating room, with standard monitoring and CVP monitoring (the central venous catheter line was also connected to the transducer), anesthesia was induced using atropine (0.2 mg), sufentanil (0.8 µg/kg), etomidate (0.2–0.3 mg/kg), and cis-atracurium (0.3 mg/kg). Radial arterial puncture and subcostal transversal fascia block were performed by the attending anesthetist after anesthesia induction and stabilization. The arterial line was connected to the Vigileo/FloTrac monitor (Edwards Lifesciences, Irvine, CA, US) for real-time hemodynamic monitoring of arterial pressure, cardiac index (CI), cardiac output (CO), and systemic vascular resistance index (SVRI). Anesthesia was maintained with propofol, remifentanil, cis-atracurium, and sevoflurane (0.5–1%). Mechanical ventilation was started with a fraction of inspired oxygen (FiO2) of 50% and a tidal volume of 8 mL/kg at a frequency of 12/min. Minute ventilation was adjusted to maintain the end–tidal carbon dioxide partial pressure (PetCO2) between 35 and 50 mmHg. Throughout the surgery, the anesthesiologist regulated the anesthesia to maintain the bispectral index (BIS) of anesthetic depth between 40 and 60.

After inducing anesthesia in both groups, fluid administration was restricted with an infusion rate of 2–3 mL/kg/h before transecting the liver parenchyma. Either milrinone or nitroglycerin was administered once pneumoperitoneum was established. In the milrinone group, 10 µg/kg milrinone was administered over 10 min as a loading dose, followed by its infusion at a rate of 0.2–0.5 µg/kg/min until the removal of the liver lesions. In the nitroglycerin group, nitroglycerin was administered at a rate of 0.2–0.5 µg/kg/min. Either milrinone or nitroglycerin infusion was stopped after complete parenchymal transection. In both groups, performing the intermittent Pringle maneuver to reduce bleeding was at the discretion of the surgeon and was used in most patients. The intermittent Pringle maneuver was performed in cycles of 15/5 min for clamping/unclamping using an appropriately prepared laparoscopic Foley catheter of the hepatic pedicle. The fluid infusion rate during liver parenchymal transection was adjusted by the attending anesthesiologist according to the CVP and surgical field condition. The surgical field condition during liver parenchymal transection was evaluated using a four-point grading scale [9, 12]: grade I is very lax hepatic veins, minimal bleeding at the resection plane, and very easy to operate; grade II is lax hepatic veins, a little bleeding at the resection plane, and easy to operate; grade III is tense hepatic veins, appreciable bleeding at the resection plane, and somewhat difficult to operate; and grade IV is very tense hepatic veins, profuse bleeding at the resection plane, and very difficult to operate. After completing parenchymal transection, the rate of intravenous fluid administration was increased to maintain the CVP close to baseline before induction. Fluids were administered intraoperatively using the lactate Ringer solution routinely, whereas the colloid solution was administered for volume replacement only when blood loss was more than 500 mL. Moreover, packed concentrated red blood cells were transfused if the intraoperative red blood cell volume detected by a blood gas analysis was lower than 25%. The total volume of intraoperative blood loss was the sum of the volumes of blood present in the suction systems and gauzes.

During the surgery, the case report form was manually recorded by an anesthetic assistant according to the device data. The hemodynamic data (heart rate [HR], pulse oximetry (SpO2), invasive MAP, CVP, BIS, PetCO2, CI, SVRI, and CO) were recorded at the following predefined time points: before (T0) and after induction (T1), before milrinone/nitroglycerin administration (T2), 10 min after milrinone/nitroglycerin administration (T3), right before the initiation of liver parenchymal transection (T4), midway through liver parenchymal transection (T5), right after the completion of parenchymal transection (T6), and at the end of the surgery (T7). Noradrenaline was administered to the patients in the two groups to maintain an MAP of more than 65 mmHg. In both groups, perioperative hypotension was defined as a 30% reduction of MAP from baseline.

Blood samples were collected to identify end-organ perfusion markers (lactate concentration) and examine routine blood factors at the beginning and end of the surgery. After the surgery, the patients woke up, were extubated in the operating room, and then were transferred to the post-anesthesia care unit with an intravenous patient-controlled analgesia device. The blood samples were taken for liver and renal function tests preoperatively, immediately after the surgery, and on postoperative days 1, 3, and 7. The diagnosis of liver cirrhosis was confirmed by the liver computed tomography and pathologic examination of the resected specimen. Additionally, we monitored postoperative recovery and complications. Postoperative complications were classified according to the Clavien–Dindo classification [13]. Postoperative mortality was defined as any death within 30 days after the surgery.

The primary outcome was the difference in blood loss between the milrinone group and the nitroglycerin group during the surgery. The secondary outcomes included the surgical field condition during liver resection, hemodynamic stability, fluid and urine volumes, perfusion parameters (intraoperative lactate concentration), postoperative liver function indicators (alanine transaminase [ALT] and aspartate aminotransferase [AST]), postoperative renal function indicators (blood urea nitrogen [BUN] and serum creatinine [Cr]), and postoperative complications.

The predefined primary endpoint was the difference in blood loss between the milrinone group and the nitroglycerin group during the surgery. Compared with the blood loss in patients in the milrinone and nitroglycerin groups undergoing hepatectomy surgery in a previous study, the mean (standard deviation [SD]) blood loss during liver resection was 240.83 (341.50) in the milrinone group and 499.23 (623.86) in the nitroglycerin group [10]. Assuming a type I error of 0.05 and a type II error β of 0.80 and rate of fall off 10%, calculations showed that 34 patients should be included in each group in order to determine a clinically relevant difference (G-power program, two-tailed). Therefore, the total sample size was 68 patients.

Data were analyzed using IBM SPSS statistical software, version 20 (IBM, Chicago, IL, US). Categorical data were presented as numbers (percentages) and compared using the chi-square test or Fisher’s exact tests, as appropriate. Continuous data were presented as the mean (SD) and compared using Student’s t-test in the case of homogeneity of variance or presented as the median (interquartile range) and compared using the Mann–Whitney U test in the case of heterogeneity of variance. Repeated-measures analysis of variance was used for intragroup comparisons at different time points. The Kappa test was used to evaluate the consistency of surgical field grading between the surgeon and the first assistant. The Mann–Whitney rank sum test was used to evaluate the difference in surgical visual field grading between the two groups. A P-value of < 0.05 was considered statistically significant for all analyses.