This study was approved by the Tianjin Medical University General Hospital Ethic Committee (Tianjin, China; Approval Number: IRB 2017-009-01), and the study protocol was registered (www.clinicaltrials.gov; Identifier: NCT03096730), and written informed consent was obtained from all patients.

Patients were randomly allocated to one of five groups: Patients in Group S receiving intraoperative sufentanil 0.30 μg kg⁻¹ and pre-anesthesia placebo (normal saline, similar volume of nalmefene); patients in Group R receiving intraoperative remifentanil 0.30 μg kg⁻¹ min⁻¹ and pre-anesthesia placebo; patients in Group N receiving intraoperative remifentanil 0.30 μg kg⁻¹ min⁻¹ and pre-anesthesia nalmefene (Tiantaishan Medicine Co., Chengdu, China) 0.20 μg kg⁻¹; patients in Group D receiving intraoperative remifentanil 0.30 μg kg⁻¹ min⁻¹ and pre-anesthesia dexmedetomidine (Enhua Medicine Co., Jiangsu, China) 0.50 μg kg⁻¹; patients in Group DN receiving intraoperative remifentanil 0.30 μg kg⁻¹ min⁻¹ and pre-anesthesia nalmefene 0.10 μg kg⁻¹ combined with dexmedetomidine 0.25 μg kg⁻¹. Nalmefene, dexmedetomidine and placebo were slowly given for 10 min before the induction of anesthesia.

All patient assignments were guided by a computer-generated random number system and individually sealed envelope. Study medication were provided by the hospital pharmacy and administered by an anesthesiologist not involved in the intraoperative management and data collection. Patients were blinded to the group allocation. Primary and secondary outcomes were assessed by another anesthesiologist responsible for the data collection, but not directly involved in the treatment of the patients and who was blinded to randomization.

Patients undergoing elective laparoscopic gynecological surgery were screened and enrolled between 4 April 2017 and 20 December 2017. Inclusion criteria were patients aged 20–64 years; American Society of Anesthesiologists physical status I or II; cognitive capacity to use the patient-controlled analgesia (PCA). The exclusion criteria were as follows: bronchial asthma; coronary heart disease; severe hypertension; diabetes mellitus; obesity (BMI >30 kg/m²); cardiac, hepatic, and renal dysfunction; psychiatric disease; history of chronic pain; history of alcohol or opioid abuse; chronic use of opioids; intake of any analgesic within 48 h before surgery; pregnancy; allergy and contraindication to dexmedetomidine or nalmefene; contraindication for the use of PCA; or incapacity to comprehend pain assessment. After randomization and allocation, patients were withdrawn if laparoscopy was converted to open surgery or if protocol was violation.

All surgical procedures were performed by senior surgeons. Patients fasted preoperatively. Upon arrival at the operating room, the patients were generally monitored by non-invasive blood pressure, ECG, heart rate (HR), pulse oximetry, and bispectral index (BIS). A peripheral intravenous line in the left arm and urinary catheter were attached before induction of anesthesia.

Induction was performed with midazolam 0.05 mg kg⁻¹, sufentanil 0.20 μg kg⁻¹, and etomidate 0.30 mg kg⁻¹, and tracheal intubation was facilitated with rocuronium 0.70 mg kg⁻¹. After intubation, all the patients were mechanically ventilated [end-tidal carbon dioxide values of 35–45 mmHg]. Anesthesia was maintained with sufentanil injection or continuous remifentanil (RenFu Co., Hubei, China) infusion as an intraoperative analgesic, and desflurane (Baxter Co., Shanghai, China) as an initial 1.3 minimal alveolar concentration (MAC) and oxygen–air mixture (fraction of oxygen, 50%). The depth of anesthesia was adjusted during surgery by 1% stepwise titration of desflurane, based on targeting BIS (45–60) and hemodynamic changes: HR exceeding pre-induction values by 15% and mean arterial blood pressure (MAP) exceeding baseline values by 20% or <60 mmHg for at least 1 min. Rocuronium (0.30 mg kg⁻¹) was administered intermittently i. v. during anesthesia. If bradycardia (HR < 45 beats·min⁻¹) and continuous hypotension (MAP <60 mmHg) persisted, additional fluid infusion, atropine (0.50 mg), and phenylephrine (0.10 mg) were also administered. During skin closure, desflurane and remifentanil were stopped, and tropisetron (2 mg) was injected. Residual neuromuscular block was antagonized by neostigmine 0.04 mg kg⁻¹ and atropine 0.01 mg kg⁻¹ when the tidal volume of spontaneous breathing exceeded 200 ml. When BIS value reached 80, response to oral command was observed, followed by eye opening and spontaneous breathing rate exceeding 10 bpm, the patient was extubated and moved to the postanesthetic care unit (PACU) for recovery at least 1 h.

RIH was characterized by the significant reduction in pain threshold to the mechanical stimuli on the dominant inner forearm (primary outcome) at 24 h after surgeries and remifentanil infusion as compared to baseline. Hyperalgesia area, pain threshold around the incision, pain intensity, cumulative sufentanyl consumption, and side-effects were the secondary outcomes investigated for 24 h after surgery.

The mechanical pain threshold was assessed using 20 hand-held von Frey filaments (North Coast Medical Inc., Gilroy, CA, United States) in an area 2–5 cm around the incision at 12 predefined positions in all four directions and on the dominant inner forearm according to our previous reports (Zhang et al., 2016). The mechanical hyperalgesia threshold was defined as the smallest force (in grams) necessary to bend a von Frey filament that was detected as painful by the patient (Bornemann-Cimenti et al., 2012; Fechner et al., 2013; Zhang et al., 2016). Also, the normalized area of hyperalgesia around the incision was measured at 24 h after surgery as previously described (Bornemann-Cimenti et al., 2012; Fechner et al., 2013; Zhang et al., 2016).

Pain intensity was evaluated on an 11-point numerical rating scale (NRS): 0 = no pain; and 10 = worst pain imaginable. The NRS score for pain at rest and after movement was assessed at 1, 3, 6, 12, and 24 h after surgery. Movement was specified as active mobilization and weight bearing while escaping any harm (Zhang et al., 2016; Zhang et al., 2017). First postoperative pain (NRS >4) was primarily managed by sufentanil titration, which was administered in 3 µg doses at intervals of 3 min until NRS <4 (Zhang et al., 2016; Zhang et al., 2017). However, sufentanil titration was discontinued if the Ramsay score (1 = anxious and agitated or restless, or both; 2 = cooperative, oriented, and tranquil; 3 = responds to command only; 4 = asleep, but has a brisk response to light tactile stimulus or a simple verbal command; 5 = asleep, but arousable only by strong physical stimulus; and 6 = asleep, unarousable) was >3, peripheral oxygen saturation decreased <92%, or breathing rate was <10 bpm. The time and total dose of first postoperative sufentanil were documented in the PACU. Furthermore, each patient was administered analgesics using a PCA pump containing sufentanil (100 µg) diluted by normal saline to a total volume of 100 ml after discharge from the PACU. The device was set to deliver a basal infusion of 2 ml h⁻¹ and bolus doses of .5 ml with a 15 min lockout period. Sufentanil comsumption was recorded at 1, 3, 6, 12, and 24 h after surgery. The incidence of postoperative side-effects was monitored during the 24 h after surgery.

A pilot study was conducted and a power analysis was implemented to calculate the sample size. The mean mechanical pain threshold of the dominant inner forearm before surgery was 96.0 g, whereas the means of the five treatment groups (Group S, Group R, Group N, Group D, and Group DN) at 24 h after surgery were 85.2, 61.5, 67.2, 73.0, and 74.0 g, respectively. We determined a difference of at least 30% (error standard deviation = 26.0) among groups. An a priori algorithm was used to estimate the required sample size for analysis of variance (ANOVA) with repeated measures. A sample size of 27 patients per group was found to be sufficient to detect a significant difference (α = 5%) with a statistical power (β-value) of 0.8. Presuming a 10% failure rate, we considered increasing the sample size to 30 patients per group.

The Shapiro–Wilk test was used to determine the normality of distribution of the data, and parametric statistics were applied. Homogeneity of variance was verified by the Levene test. Data from the NRS scores, and mechanical pain threshold were analyzed by two-way repeated measures with Bonferroni post hoc comparisons. Data from time and total dose of first postoperative sufentanil titration, sufentanil consumption by PCA, and normalized area were analyzed by one-way with Bonferroni post hoc comparisons. Other quantitative data, such as age, weight, duration of remifentanil infusion, mean concentration of desflurane, recovery time and extubation time were also analyzed using one-way with Bonferroni post hoc comparisons. Simultaneously, the χ² test and Fisher’s exact test were used to analyze categorical variables, such as atropine administration, hypotension, dizziness, nausea, and vomiting. Data were expressed as the mean (SD) or the number of patients/percentage. A p-value <0.05 was considered statistically significant. SPSS 21.0 software (SPSS, Inc., Chicago, IL, Unitred States) was used for all statistical analysis.

Among the one hundred and fifty-eight patients recruited, one hundred and fifty patients were eligible for inclusion. Eleven patients were withdrawn after protocol violation or conversion to open surgery, and one hundred and thirty-nine patients completed the study (Figure 1). The five groups were balanced in terms of patient characteristics (Table 1).

No significant difference was detected between groups with respect to duration of surgery, duration of remifentanil infusion, mean volume of desflurane, intraoperative MAP and HR, and the proportion of patients administered phenylephrine (Table 1; Figure 2). However, Group D exhibited the higher proportion of patients administered atropine as compared to Group R (p = 0.044, Table 1). When compared with Group R, Group S exhibited delayed eye opening time [7.1 (1.7) vs. 8.7 (2.2) min, p = 0.009] and extubation time [8.4 (1.6) vs. 10.2 (2.3) min, p = 0.005]. Simultaneously, patients in Group D exhibited delayed extubation time as compared to patients with remifentanil infusion alone [10.0 (1.7) vs. 8.4 (1.6) min, p = 0.012]. However, eye opening time and extubation time did not differ between Groups R, N and DN.

As shown in Figures 3A, B, the baseline mechanical pain threshold on the dominant inner forearm and around the incision was similar among all the groups (p > 0.05). Compared with baseline, intraoperative remifentanil infusion induced a robust decrease in pain threshold on the forearm and peri-incisional area at 6 h (p < 0.001 and p < 0.001) and 24 h (p < 0.001 and p < 0.001) after surgery. Interestingly, as compared to patients in Group R, higher levels of pain threshold on the forearm and around the incision were detected in Group D at postoperative 6 h (p = 0.002 and p < 0.001) and 24 h (p = 0.01 and p = 0.03), and Group DN at postoperative 6 h (p = 0.003 and p < 0.001) and 24 h (p < 0.001 and p = .002). Also, patients in Group N showed the higher threshold on the forearm at postoperative 24 h (p = 0.039) than that in Group R. However, the threshold around the incision was comparable in Groups R and N at 6 h (p = 0.14) and 24 h (p = 0.075) after surgery. Besides, similar threshold levels at postoperative 24 h were seen on the forearm (p = 0.73) and peri-incisional site (p = 0.97) in Groups D and DN.

As shown in Figure 3C, the normalized area of hyperalgesia around the incision in Group R was larger than that in Group S at postoperative 6 h (p < 0.001) and 24 h (p = 0.001). As compared to Group R, it was downregulated in Groups D and DN at postoperative 6 h (p = .03 and p = 0.007) and 24 h (p = 0.015 and p = 0.009). However, there was no significant difference between Group R and Group N at postoperative 6 h (p = 0.22) and 24 h (p = 0.74).

As shown in Figure 4, NRS scores at rest were higher in Group R than Group S at 3 h (p = 0.009), 6 h (p = 0.019), 12 h (p = 0.004) and 24 h (p = 0.004) after surgery. Similarly, patients in Group R showed the higher NRS scores after movement at postoperative 3 h (p = 0.003), 6 h (p = 0.004), 12 h (p = 0.007) and 24 h (p = 0.013) as compared to Group S. Strikingly, compared with Group R, Groups D exhibited the lower NRS scores at rest at postoperative 3 h (p = 0.042) and 12 h (p = 0.039) and NRS scores after movement at 3 h (p = 0.007) and 24 h (p = 0.046) after surgery. Furthermore, NRS scores at rest were lower in Group DN than Group R at 12 h (p = 0.004) after surgery. NRS scores after movement were lower in Group DN than Group R at 3 h (p = 0.025), 12 h (p = 0.046) and 24 h (p = 0.026) after surgery. However, there was no significant difference in NRS scores between Group R and Group N (p > 0.05).

As compared to Group S [25.0(7.3) min], patients in Group R [21.4(5.9) min] showed the shorter time to first postoperative sufentanil requirement in the PACU (p = 0.001). When compared with Group R, it was prolonged in Group D [24.8(6.3) min, p = 0.003] and Group DN [24.4(6.1) min, p = 0.013]. However, no significant difference was detected between Group R and Group N [23.7(7.9) min, p = 0.133].

Table 2 displays postoperative analgesics administration. As compared to Group R, the demand of sufentanil titration in Group S (p < 0.001), Group D (p = 0.036) and Group DN (p = 0.025) was significantly decreased, whereas there was no significant difference between Group R and Group N (p = 0.474). Furthermore, sufentanil consumption by PCA in Group S was significantly less than that in Group R. More importantly, when compared with Group R, sufentanil requirement was greater in Group D during the second 6 h (p = 0.007) and the third 6 h (p = 0.02) after PCA and in Group DN during the third 6 h (p = 0.012) and the fourth 6 h (p = 0.036) after PCA. However, no difference was reported between Group R and Group N.

The incidence of postoperative nausea in Group S were higher than in Group R (p = 0.044, Table 3), whereas no differences were found among Groups R, N, D, and DN (p > 0.05, Table 3). Additionally, the incidence in postoperative hypotension, bradycardia, vomiting, shivering, somnolence, dizziness and respiratory depression didn’t differ among the five groups (p > 0.05, Table 3).