Adenosine for postoperative analgesia: A systematic review and meta-analysis

Purpose Perioperative infusion of adenosine has been suggested to reduce the requirement for inhalation anesthetics, without causing serious adverse effects in humans. We conducted a meta-analysis of randomized controlled trials evaluating the effect of adenosine on postoperative analgesia. Methods We retrieved articles in computerized searches of Scopus, Web of Science, PubMed, EMBASE, and Cochrane Library databases, up to July 2016. We used adenosine, postoperative analgesia, and postoperative pain(s) as key words, with humans, RCT, and CCT as filters. Data of eligible studies were extracted, which included pain scores, cumulative opioid consumption, adverse reactions, and vital signs. Overall incidence rates, relative risk (RR), and 95% confidence intervals (CI) were calculated employing fixed-effects or random-effects models, depending on the heterogeneity of the included trials. Results In total, 757 patients from 9 studies were included. The overall effect of adenosine on postoperative VAS/VRS scores and postoperative opioid consumption was not significantly different from that of controls (P >0.1). The occurrence of PONV and pruritus was not statistically significantly different between an adenosine and nonremifentanil subgroup (P >0.1), but the rate of PONV occurrence was greater in the remifentanil subgroup (P <0.01). Time to first postoperative analgesic requirement in the adenosine group was not significantly difference from that of the saline group (SMD = 0.07, 95%CI: −0.28 to 0.41, P = 0.71); but this occurred significantly later than with remifentanil (SMD = 1.10, 95%CI: 2.48 to 4.06, P < 0.01). Time to hospital discharge was not significantly different between the control and adenosine groups (P = 0.78). The perioperative systolic blood pressure was significantly lower in the adenosine than in the control group in the mannitol subgroup (P < 0.01). The incidence of bradycardia, transient first- degree atrioventricular block, and tachycardia was not significantly different between the adenosine and control groups (P > 0.1). Conclusion Adenosine has no analgesic effect or prophylactic effect against PONV, but reduce systolic blood pressure and heart rates. Adenosine may benefit patients with hypertension, ischemic heart disease, and tachyarrhythmia, thereby improving cardiac function.


Results
In total, 757 patients from 9 studies were included. The overall effect of adenosine on postoperative VAS/VRS scores and postoperative opioid consumption was not significantly different from that of controls (P >0.1). The occurrence of PONV and pruritus was not statistically significantly different between an adenosine and nonremifentanil subgroup (P >0.1), but the rate of PONV occurrence was greater in the remifentanil subgroup (P <0.01). Time to first postoperative analgesic requirement in the adenosine group was not significantly difference from that of the saline group (SMD = 0.07, 95%CI: −0.28 to 0.41, P = 0.71); but this occurred significantly later than with remifentanil (SMD = 1.10, 95%CI: 2.48 to 4.06, P < 0.01). Time to hospital discharge was not significantly different between the control and adenosine groups (P = 0.78). The perioperative systolic blood pressure was significantly lower in the adenosine than in the control group in the mannitol subgroup (P < 0.01). The incidence of bradycardia, transient first-degree atrioventricular block, and tachycardia was not significantly different between the adenosine and control groups (P > 0.1). a1111111111 a1111111111 a1111111111 a1111111111 a1111111111

Introduction
Postoperative pain can have a significant effect on patient recovery. The results of 1 survey have indicated that about 80% of patients who undergo surgery experience severe postoperative pain [1].Preemptive analgesia has therefore been widely used in the clinical setting; opioids remain the mainstay for postoperative analgesia, especially after major surgery [2]. However, opioid-related side-effects can be distressing to patients, and include respiratory depression, post-operative nausea and vomiting (PONV), urinary retention, sedation, and pruritus [3].To improve analgesic quality, adjunct medications are used; such multimodal analgesia allows more optimal pain management [4].
Adenosine(ADO) is a breakdown product of adenosine triphosphate(ATP); it is an endogenous purine nucleoside with several biological effects. It is involved in numerous biological processes, including neurotransmission, muscle contraction, heart function, and has antiinflammatory activity [5]. Perioperative infusion of low-dose adenosine in recent studies has been found to reduce the requirements for inhalation anesthetics [6]. It could be argued that the vasodilatory effect of ADO has a major influence on the stability of systolic blood pressure (SBP) during surgery. 2 groups with similar presurgical ET-IS0 concentrations indicated the absence of significant ADO-induced reduction of SBP. This further implies that the maintained SBP reported in the ADO group during surgery is unrelated to an ADO-mediated vascular effect, but is rather due to an ADO-mediated modulation of the afferent reflex response to surgical trauma, which is related to the antinociceptive action of adenosine [6]To date, no serious adverse effects of ADO treatment have been reported in humans [6,7].
Here, we performed a meta-analysis to evaluate the impact of perioperative administration of ADO on postoperative pain, PONV, and the cardiovascular system.

Methods
We followed the recommendations of the PRISMA statement [8] Search strategy Published reports [2,6,7,9-16,] in English were retrieved in a computerized search of Scopus, Web of Science, PubMed, EMBASE, and Cochrane Library databases, for the period up to July 2016. The keywords used to search these databases were adenosine, postoperative analgesia and postoperative pain(s), and humans, RCT, and CCT were used as filters. We also examined similar published research with retrieved literatures.

Study selection
The details of the study selection processes are shown in Fig 1. The database search yielded 134 potentially relevant papers. However, 121 studies were excluded after reviewing the title and abstract as these included duplicated and non-original studies, while two were excluded as they did not meet the inclusion criteria; 11 studies were thus further investigated [2,6,7,[9][10][11][12][13][14][15][16]. remaining 4 studies did so via the intrathecal route [10,12,16]. The characteristics of these studies are summarized in Table 1, and risk of bias in these studies is summarized in Table 2.

Primary outcomes
Pain scores at 4 h after surgery. The overall effect of ADO on pain scores at 4 h after surgery, as compared with controls, showed no significant difference (SMD = -0.08, 95%CI: -0.26 to 0.10, I 2 = 53%, P = 0.37),according to the forest plot (Fig 2). The funnel plot showed significant asymmetry (P < 0.05; Fig 3).
Sex subgroup analysis Six studies included only female patients(women subgroup) [2,6,7,9,11,13].As the study by Ghai et al. [13]did not include the standard deviation for VAS pain scoring, it had to be excluded for the sake of accuracy. Compared with the ADO group, Sex in the women subgroup (SMD = −0.03, 95%CI: −0.20 to 0.26, I 2 = 0%, P = 0.78) was not statistically significantly different from a combined man/women subgroup (Fig 2).
Pain scores at 24 h after surgery. According to the overall forest plot for the effect of ADO on pain scores at 24 h after surgery, VAS/VRS scores in the treatment group was not significantly different from that in the control group (SMD = −0.03, 95%CI: −0.33 to 0.28, I 2 = 51%,P = 0.86; Fig 4). As only 6 studies investigated this time-point, the funnel plot was not suitable for evaluating publication bias.
Pain scores at 48 h after surgery. Only 1 study [2]examined the pain scores at 48 h after surgery s. In that study, there was no significant difference between the control group (a saline group) and the ADO group (P = 0.14).
Cumulative postoperative opioid consumption. The overall effect of ADO on early pain in terms of postoperative opioid consumption was not significantly different from that of control treatments (SMD = −0.74, 95%CI: −2.17 to 0.69,I 2 = 92%, P = 0.31) according to the forest  plot ( Fig 5). As this was recorded by only 6 studies, the funnel plot was not suitable for evaluating publication bias.
Cumulative opioid consumption at 4h postoperatively In terms of cumulative opioid consumption at 4 h, 3 studies were included: 2 for the nonremi subgroup [6,10] and 1 for the remifentanil subgroup [14].We show the cumulative opioid consumption at 4 h separately for these subgroups. In the nonremi subgroup, outcomes of the control groups were not significantly different from those of the ADO group (SMD = 0.04, 95%CI: −0.41 to 0.48, I 2 = 0%, P = 0.88); in the remifentanil subgroup, the control group used significantly more opioids than did the ADO group (SMD = −2.35, 95%CI: −3.16 to −1.54, P < 0.00001).
Cumulativeopioid consumption at 48h postoperatively Two studies [2,14] were included in the analysis of the cumulative opioid consumption at 48h. This value was significantly more in the control than in the ADO group (SMD = −6.84, 95%CI: −14.72 to1.04, I 2 = 93%, P < 0.09, Fig 8). The heterogeneity in the data was higher, as one study [2] used saline in the controls (SMD = 0, 95%CI: −8.68 to 8.68); but the other study [17] used remifentanil in the controls (SMD = −39.00, 95%CI: −57.81 to −20.19). In the study by Fukunaga et al. [14], the usage of cumulative opioid consumption at 48 h was less in the ADO group than in the controls.

Secondary outcomes
Postoperative nausea and vomiting. Four studies examined the effect of ADO on PONV [2,6,13,14]. In 4 of these studies studies, except for that by Ghai et al. [13], nausea was combined with vomiting as a single outcome, while the other studies [2,8,14] considered them as separate outcomes.
Postoperative nausea Four studies were included in the analysis of nausea [2,6,13,14]. The effect in the remifentanil subgroup suggested that postoperative nausea occurs significant more frequently in the control than in the ADO group (OR = 0.00 [0.00 to 0.07], P < 0.01; however, in the nonremi subgroup, there was no significant difference between the control and ADO group(P = 0.26; Fig 9).
Postoperative vomiting Two studies included an analysis of postoperative vomiting [2 14]. The effect in the study by Fukunaga et al. (remifentanil subgroup) [14] suggested that this occurred significantly more frequently in the control than in the ADO group (OR = 0.03 [0.00   to 0.50]), but there was no significant difference between controls and the ADO group in the nonremi subgroup (Fig 10).
Pruritus. Only 1 study [2] analyzed the occurrence of pruritus; there was no significant difference between the control and ADO group (P = 0.54).   [2,16] and 1 study in the remifentanil subgroup [15]. In the saline subgroup, this outcome was not significantly different in the control compared to the ADO group (SMD = 0.07, 95%CI: −0.28 to 0.41, I 2 = 0%, P = 0.71); in the remifentanil subgroup, this occurred significantly earlier in the control than in the ADO group (SMD = 1.10, 95%CI: 2.48 to 4.06,P <0.01; Fig 11).
Time to hospital discharge. Only the study by Habib et al. [2] observed the effect of ADO in time for hospital discharge. They found no significant difference between the control group with the ADO group(P = 0.78). Effects on cardiovascular system. We assessed the overall effects of ADO on the cardiovascular system, as compared with the control and treatment; several aspects were considered, as follows: systolic blood pressures (SBp), diastolic blood pressures (DBp), heart rate (HR), bradycardia, tachycardia, and transient first -degree atrio-ventricular block (AVB.) Systolic and diastolic blood pressures Both SBP and DBP at different surgical stages were evaluated to determine the effects of ADO on blood pressure. The surgical processes were divided into 3 parts: early surgical period, mid-surgical period, and late surgical period.;. For assessing SBP, 4 studies were eligible;3 studies were includedin mannitol subgroup [6,7,9], and only1study [10] was included in the IT+ saline subgroup. The early SBP was significantly lower in the ADO than the control group in the mannitol subgroup (SMD = −19.26, 95%CI:−21.37 to −17.14, I 2 = 88%, P <0.01), while early SBP was similar between theA-DOand the control group in the IT+saline subgroup (SMD = 7.00, 95%CI: −0.95 to 14.95, P = 0.08, Fig 12).
For SBP in the late surgical period, the studies included and the trends of the outcome were almost the same as for the early SBP (Fig 14).
For DBP, only 1 study [2] could be assessed. The study [2] was included in the IV+ saline subgroup. The DBP in the ADO group was significant lower than that in the controls (SMD = −17.00, 95%CI: −21.91 to −12.09, P <0.01).
One study [18] was included in the analysis of hypotension; this study used remifentanil as the control. There was no significant differences between the ADO and control groups (P = 0.15).
Heart rate We also considered the effect of ADO on HR in to 3 parts: early surgical period, mid-surgical period, and late surgical period.separately. For early HR,3 studies were eligible;2 studies were included in the mannitol subgroup [6,7],and only 1 study [10] was included in the IT+saline subgroup. The early HR in the ADO group was not significantly different from that in the control group in the mannitol (P = 0.20) of IT+ saline subgroups (P = 0.10; Fig 15).
For evaluating mid HR, we included 4 studies. 2 studies were included in the mannitol subgroup [6,7], and 1 study [2] was included in the IV+ saline subgroup and one study [10] in the IT+ saline subgroup. The mid HR in the ADO group was not statistically significantly different from that in the control group in the mannitol(P = 0.74)and IT+ saline subgroups (P = 0.27); Adenosine for postoperative analgesia however, the mid HR was higher in the ADO group than in the control group in the IV+ saline subgroup (P = 0.002; Fig 16).
For assessing HR in late surgical period, 3 studies were included: 2 studies were included in the mannitol subgroup [6,7],while 1 study [10] was included in the IT+ saline subgroup. The HR was significant higher in the ADO group than in the control group in the mannitol subgroup (SMD = 4.06, 95%CI:2.37 to 5.74, P <0.00001). However, the late HR in the IT+ saline subgroup was not significantly different between the treatment and control groups (P = 0.12; Fig 17).
For both bradycardia and transient first degree AVB, two studies [2,15] were eligible for inclusion. 1 study [2] was included in the IV+ saline subgroup, but the study of Lee et al. [15] used remifentanil in the controls. The incidence of bradycardia (P = 0.24) and transient first degree AVB (P = 0.17) was not significantly different in the ADO than in the control groups (Figs 18 and 19).For tachycardia evaluation, 1 study [2] was in it. However, the comparison between ADO and control was not significantly different.

Discussion
This meta-analysis suggested that ADO administration did not change postoperative VAS/ VRS scores by 24 h, irrespective of sex, or whether it was administered by IT or IV, or with mannitol. In comparison with remifentanil, ADO reduced postoperative pain by 24 h, and reduced opioid consumption over the first 48 h; however, in comparison with other non-remifentanil control treatments, ADO had no significant effect on pain score or opioid consumption. Moreover, the time to first analgesic requirement was much earlier in remifentaniltreated than in the ADO group, but ADO treatment did not differ from saline treatment in this respect. The occurrence of PONV was higher in the remifentanil-treated group than in the ADO-treated group. Previous studies [18,19] had suggested that the occurrence of PONV was associated with an increased usage of opioids. Nevertheless, our study showed no statistically significant difference between ADO and saline treatment. SBP and HR was lower in the ADO treated group than in the control group in the IV+ saline/mannitol subgroups throughout surgery, but in the treatment and control groups did not differ in the IT+ saline subgroup. The occurrence of bradycardia and first degree AVB did not differ statistically significantly between ADO and remifentanil treated groups.
The study of Chiari et al. [20] verified the safety of ADO, and neurological function unaffected in histopathological animal studies. In many studies, ADO has been used as ananalgesic mediator in animal and human experiments [21][22][23][24][25].The present meta-analysis also showed that exogenous ADO relieved postoperative pain by 24h, thereby reducing opioid requirements. These results were in accordance with the studies by Katz et al. and Martins et al. [26,27] in that endogenous ADO production was found to improve mechanical hyperalgesia. The first postoperative analgesic requirement was about 25 times longer than in the remifentanil subgroup. This effect was thought to be mediated throughout the nervous system by the ADO A1 receptor (A1R) [28]. ADO modulates signal transmission of pain in the periphery and the spinal cord. A1Rs can change the extracellular availability of ADO and subsequently regulate pain transmission. While the study of Sjolund et al. [25] suggested that the pain-reducing effect of ADO is due to a reduction of substance P (SP) in cerebrospinal fluid. It has been suggested that SP facilitates the excitatory amino acid-induced activation of the N-methyl D-aspartate (NMDA) via NK-1 receptor stimulation. The proposed role for SP is supported by the results of behavioral studies in which SP antagonists were used in animal models of peripheral inflammation. The study of Omoigui et al. [29] proposed that the origin of all pain is inflammation and the inflammatory response. However, the effect of ADO was only found in comparison with remifentanil and ADO, but not with saline.
The ORs of PONV in the ADO group was much lower than in the remifentanil subgroups. A study by Murataet al. [30] proposed that ADO serves to counteract further progression of hyperemesis gravidarum. Therefore, ADO seems to have a positive protective action against nausea and vomiting. In this meta-analysis, the effect was only seen in comparison between remifentanil and ADO, but not between saline and ADO.
Moreover, we found that SBP and HR of participants treated with ADO in the IV+ mannitol/saline subgroups were lower than those of the controls throughout the surgery. Thus, ADO administered by IV had marked effects on HR and BP. These effects are probably exerted through theA1, A2A, A2B, and A3 ADO receptor subtypes, which are all expressed in myocardial cells [31 32]. Activation of both the A1Rs and A2Rs is involved in the regulation of HR, whereas A3Rs play a key role in cardioprotection [32][33][34]. These effects reflected the classic secondary action of ADO.
Thus, ADO had the same impact as saline in terms of analgesia, but reduced heart rate and blood pressure in our meta-analysis. However, the rate of occurrence was not statistically significantly different between ADO and saline groups in terms of bradycardia and transient first-degree AVB. This may be because ADO induces the baroreceptor reflex response to hypotension and directly stimulates the sympathetic nervous system regardless of changes in blood pressure [35].In the study by Takeshi et al. [36], almost 30% of patients demonstrated reduced heart rates after ADO infusion. This could be influenced by smoking, using the βblockers, higher resting HR, lower ejection fraction, etc., both smoking and β-blockers might destroy the pharmacodynamics of ADO. However, higher resting HR and lower ejection fraction could be associated with the baroreceptor reflex response [35]. The studies included in our meta-analysis did not contain any information on smoking, β-blocker use, or ejection fraction. This should be investigated further in future.
Additionally, ADO is a powerful cardioprotective mediator in ischemic preconditioning. Any factors that are liable to increase ADO accumulation in the heart during myocardial ischemia-reperfusion would reduce myocardial injury [37]. Intracoronary ADO may be an effective therapy for no-reflow in ST-segment elevation myocardial infarction [38]. A metaanalysis by Singh et al. [39] showed that intracoronary ADO administration was well-tolerated and significantly improved electrocardiographic outcomes, with a tendency towards improvement of adverse cardiovascular events, heart failure, and cardiovascular mortality. Safety analysis showed no significant difference in chest pain events (RR 1.26, 95%CI: 0.55 to 2.86; P = 0.58), bradycardia (RR 2.19, 95%CI: 0.24 to 0.38; P = 0.49), ventricular tachycardia (OR 0.61, 95%CI:0.08 to 4.90; P = 0.64), and ventricular fibrillation (RR 0.49, 95%CI: 0.13 to 1.90; P = 0.30), as compared with the placebo group.
Furthermore, ADO activates four receptors (A1R, A2aR, A2bR, A3R) to reduce ischemiareperfusion injury, as ADO can decrease mechanical obstruction of capillaries by neutrophils. Moreover, it can block the release of vasoconstrictors (e.g.,leukotrienes, platelet activating factor, endothelin) by activated neutrophils and platelets to protect the ischemic myocardium [40]. In the study by Ernens et al. [41], ADO could stimulate increased production of thrombospondin-1 by human macrophages. In rats, chronic ADO administration can increase border-zone vascularization of myocardial infarction lesions, via increase thrombospondin-1 expression. This effect is mediated via the cAMP/PKA pathway and involves A2AR and A2BR [41]. Hence, ADO can improve cardiac function and reduce the area of myocardial infarction in multiple protective ways.
However, our meta-analysis has some limitations. We set out to perform some sensitivity analyses and subgroup analyses to assess the impact of factors on primary outcomes. These analyses may have performance, publication, or reporting bias; and the results should be considered with due caution. The small numbers of included trials was another limitation; we could not apply funnel plots and meta-regression to examine all comparisons and ORs; because these analyses were needed at least 10 studies. The small numbers of included trials caused wide CIs and ORs.
In conclusion, this meta-analysis suggested that ADO has no analgesic effect or prophylactic effect on PONV. However, ADO can reduce SBP and participants' HR in part during surgery. It did not increase the rate of occurrence of bradycardia and first-degree AVB. This may benefit patients with hypertension, ischemic heart disease (e.g.,coronary heart disease, congestive heart failure) and tachyarrhythmia(e.g.,tachycardia caused by hyperthyroidism, supraventricular tachycardia), to improve cardiac function.