A prospective, randomized, and double-blinded investigation was performed on patients underwent elective open liver resection in Beijing You An Hospital, Capital Medical University. The study protocol was approved by the Research Ethics Committee of Beijing You An Hospital, Capital Medical University in Beijing, China [Ref: (2016)10]. The trial was registered at chictr.org at the beginning of the study (Ref: ChiCTR-IPR-17011001, March 2017), and written informed consent was obtained from all patients. Patients who undergoing elective open liver resection for malignant and benign liver lesions under general anesthesia from March 2017 to August 2017 were enrolled in this study if they met the following inclusion criteria: age between 18 and 70 years, American Society of Anesthesiologists (ASA) grade I–II. Exclusion criteria included laparoscopic liver resection, emergency operation, blood loss greater than 1000 mL, ASA physical status 3 or greater, Child-Pugh class C, allergy to any drug/agent used in the study, pre-existing chronic pain, opioid dependence, a left ventricular ejection fraction of <35%, greater than first-degree atrioventricular block, chronic renal dysfunction requiring replacement therapy, altered mental status, inability to provide adequate informed consent, or patient refusal, use of psychiatric medications, pregnant or lactating.

Before the operation, patients were randomized into a control (Con) group or a Dex group (30 per group). The randomization sequence without stratification was generated by a computer, and sealed with consecutively numbered envelopes. Treatment allocation was revealed by opening the envelope by nurses on the morning of surgery, the nurses were responsible for preparing DEX or saline placebo with the same 50 ml syringe and the post-operative patient controlled intravenous anesthetic drugs. And then the anesthesiologist administered the drug in a blind mode. Patients and investigators were all blinded to group allocation until the final statistical analysis was completed.

Oxygen saturation, electrocardiography, end-tidal CO₂, arterial pressure and temperature were continuously monitored with an automated system (Philips IntelliVue MX800) when patients arrived at the operating room, then a peripheral intravenous catheters (G14) were placed before induction. A double lumen catheter (Foshan Special Medical Co., Ltd., China) was inserted in the right internal jugular vein for central venous pressure (CVP) measurement and a radial artery cannula (Becton Dickinson Infusion Therapy Systems, United States) was inserted for arterial pressure measurement and arterial blood gas test. Surface electrodes for the bispectral index (BIS; Aspect Medical System, Norwood, MA, United States) were placed to the patient’s forehead. The urine output was monitored through a 14 Fr catheter after the induction of anesthesia. A forced-air warming device (Patient Warming System, Warm Touch, United States) was used in both groups to maintain normothermia.

Patients in Dex group received intravenous DEX (0.5 μg⋅kg^-1; Precedex; Aibeinin^®, Inc., Henrui Pharmaceutical, China) over 10 min before endotracheal intubation, then adjusted to 0.3 μg⋅kg^-1⋅h^-1 from the 50 ml syringe until incision suturing. While in the Con group, an initial loading dose of 0.9% sodium chloride was administered as placebo followed by a maintenance dose from the same syringe at a relevant rate to the Dex group. All patients were induced with the same agents: Oxycodone (0.2 mg⋅kg^-1; Hamol LIMITED, United Kingdom), propofol (2 mg⋅kg^-1), and cisatracurium (0.2 mg⋅kg^-1). After tracheal intubation, anesthesia was maintained with an infusion of propofol 4–8 mg⋅kg^-1⋅h^-1, a remifentanil (0.05–0.4 μg⋅kg^-1⋅min^-1) and cisatracurium (1–2 μg⋅kg^-1⋅min^-1) infusion according to clinical needs. Mean arterial pressure (MAP) was kept at a target of 65 mm Hg. Ringer’s lactate solution and gelofusine were used during the surgery. Positive pressure ventilation were maintained with endotracheal intubation to achieve an end-tidal pressure of CO₂ of 30–40 mmHg. The rate of remifentanil and propofol were adjusted according to the hemodynamic stability, and depth of anesthesia was evaluated using the bispectral index at a target around 50 during surgery.

The baseline heart rate (HR) and MAP were defined as the mean of the two lowest measurements recorded before the induction of anesthesia. During the operation, tachycardia and bradycardia were defined as a > 30% increase from baseline or HR < 45 beats per minute (bpm) and were treated using esmolol (10–20 mg) or atropine (0.2–0.5 mg), respectively. Hypertension was defined as a > 30% increase from baseline and was treated by increasing the remifentanil; and if hypertension persistent after remifentanil treatment, urapidil (10–15 mg) was administered. Hypotension was defined as a > 30% decrease from baseline and was treated by decreasing the remifentanil, and if persistent, by incensement the rate of fluid, administration of phenylephrine (20–50 μg) or ephedrine (6–12 mg) according to the clinical need. DEX and cisatracurium were discontinued approximately 30 min before completion of the surgery, while all patients received tropisetron (5 mg) and oxycodone (0.1 mg⋅kg^-1). Neostigmine and atropine were given to reverse residual neuromuscular block at the end of surgery. All patients were transferred to the post-anesthesia care unit (PACU) for further observation after extubation in the operating room.

One day before surgery, the patients were instructed to use the PCA pump (Apon, China) and a 10 cm Visual Analog Scale (VAS) score (0–3 mild pain; 4–6 moderate pain; 7–10 severe pain). PCA protocol in Con group (without a background infusion) consisted of 60 mg oxycodone diluted to 120 ml and administered at a bolus dose of 2 ml, with 5 min lockout interval and a 1 h limit of 20 mL, while in Dex group was 60 mg oxycodone and 360 μg Dex diluted to 120 ml with the same regimen as those in Con group after arriving at the PACU.

Patients were encouraged to press the pump button when they experienced pain with cough at a severity of VAS score > 4. If the pain control was not satisfied, then rescue analgesia (40 mg of parecoxib or 100 mg of tramadol by intravenous injection) was used. The PCIA pump will be stopped on the second post-operative day, and PONV will be treated with 5 mg of tropisetron according to the patient’s request.

Age and body mass index (BMI; calculated as weight (kg)/[height (m)]²), ASA, Child–Pugh scores and operative time were collected from the patient charts. Child–Pugh scores employs five clinical measures of liver disease. Each measure is scored 1–3, with 3 indicating most severe derangement. Chronic liver disease is classified into Child–Pugh class A to C, employing the added score from above. Class A, 5–6; Class B, 7–9; Class C, 10–15 (Cholongitas et al., 2005). Duration of surgery was measured from the surgical incision to the closure of the skin. HR and MAP were got from the Philips IntelliVue monitor at the following points: arrival at the operating room (T0), just before intubation (T1), after intubation (T2), during surgical incision (T3), on extubation (T4), on arrival at the PACU (T5), and 20 min after arriving at the PACU (T6). The requirement of narcotic and vasoactive drugs, estimated blood loss, fluid requirements, and complications were recorded. The PACU stay time was also recorded based on the Aldretes criteria (Aldrete, 1995).

The cumulative amount of self-administered oxycodone was recorded until 48 h after the surgery. Pain at rest and with cough was assessed with the VAS, where 0 denotes no pain and 10 denotes the worst pain experienced. The Ramsay Sedation Scale was also used to assess the sedation state, where 1 = anxious, agitated, or restless; 2 = cooperative, oriented, and tranquil; 3 = responds to command; 4 = asleep but with brisk response to a light glabellar tap or loud auditory stimulus; 5 = asleep, sluggish response to a light glabellar tap or loud auditory stimulus; and 6 = asleep, no response to stimuli. PONV scores from 1 to 4 (1 = without nausea and vomiting, 2 = nausea without vomiting, 3 = less than two times vomiting, 4 = severe vomiting more than two times) (Apfel et al., 1999). These were evaluated at 1, 4, 8, 12, 24, 48 h after surgery. Patients were asked to rank their satisfaction with pain control according to the following scale: 1 = not satisfied, 2 = moderately satisfied, 3 = satisfied, 4 = very satisfied at 48 h after the PCA pump was removed (Harms et al., 2007). The number of rescue analgesia and adverse effects (such as hypertension, hypotension, bradycardia, tachycardia, respiratory depression, liver failure, and death within 48 h after surgery) was also recorded.

Data are expressed as the mean ± SD, percentage (%), and numbers (n). The distribution of variables was assessed by the Kolmogorov–Smirnov. Homogeneity of variance was determined using Levene’s tests. Inter-group comparisons were performed using repeated-measures analysis of variance. The non-parametric Mann–Whitney–Wilcoxon Test was used for variables that were not normally distributed. Categorical data were analyzed using Fisher’s exact tests or chi-squared tests. Probability (P) values of <0.05 were considered statistically significant. Statistical analysis was performed using SPSS for Windows Version 21.0 (SPSS Inc., Chicago, IL, United States).

The CONSORT demographics are described in a patient enrollment flow diagram. Sixty patients who underwent liver surgery were screened from March 2017 to August 2017. Eight patients were excluded because of not meeting the inclusion criteria: five patients refused to participate, the blood loss of three patients was more than 1000 ml. Finally, 52 patients were finally enrolled in the primary analysis and randomly divided into two groups: 26 patients in Dex group and 26 patients in Con group. No patients failed to follow-up.

Baseline characteristics of the 52 patients were comparable between the groups. As shown in Table 1, there were no significant differences in age, weight, height, BMI, sex, ASA grade, Child–Pugh grade, tumor type and history of comorbidity (all P > 0.05; Table 1).

AS displayed in Figure 1, baseline HR and MAP were similar between the two groups (P > 0.05; Figure 1). Compared with Con group, HR in Dex group was statistically decreased from T1 to T6, while MAP was significantly decreased from T1 to T3 (P < 0.05; Figure 2), but MAP was similar in two groups from T4 to T6 (P > 0.05; Figure 2).

As shown in Table 1, there were no statistically differences in operating time, duration of anesthesia, bispectral index, estimate blood loss, fluids infusion, urine output, and consumption of cisatracurium (P > 0.05; Table 1). The consumption of propofol and remifentanil were significantly decreased in Dex group Compared with Con group (P = 0.000; Table 1).

The patients required for intraoperative vasoactive drugs such as esmolol (3 vs. 9, P = 0.048), and urapidil (4 vs. 11, P = 0.032) were less in Dex group than Con group. While the usage of atropine, ephedrine, and phenylephrine were similar between the two groups (Table 2).

As displayed in Figure 2, the total dosage of oxycodone was significantly lower in Dex group than in Con group at 4, 8, 12, 24, and 48 h after surgery (P < 0.05, Figure 2). The VAS scores at rest at 1, 4, and 8 h post-operative and with cough at 24, and 48 h after surgery were statistically lower in Dex group compared with Con group (P < 0.05, Figure 3), while the VAS scores at other time were similar between the two groups. There was no significant difference in RSS within the first 48 h after surgery between the two groups (P > 0.05, Figure 4).

More patients required rescue analgesia in Con group compared with Dex group within 48 h after surgery (3 VS. 9, P = 0.048, Table 1). The first exhaust time (54.04 ± 6.88 VS. 61.38 ± 11.42 min, P = 0.007) were earlier in Dex group than in Con group after surgery (Table 1). Satisfaction with pain control was statistically higher in Dex group than in Con group (P < 0.05, Figure 5A), while the incidence of PONV was less in Dex group than in Con group (P < 0.05, Figure 5B).

As shown in Table 1, PACU time and the incidence of other main adverse events such as hypertension, hypotension, tachycardia, and bradycardia are similar in two groups (P > 0.05, Table 1). No respiratory depress, liver failure, and no patients died within 48 h after surgery (Table 1).