Meta-analysis of remote ischemic conditioning in patients with acute myocardial infarction

Effects of remote ischemic conditioning (RIC) in acute myocardial infarction (AMI) patients remain conflicting. We performed this meta-analysis of randomized clinical trials (RCTs) to evaluate the benefits of the RIC in patients with AMI. Potentially relevant RCTs were identified by searching PubMed, Embase, Cochrane Library, VIP, CNKI, and Wanfang database until November 2016. RCTs evaluating RIC using intermittent limb ischemia-reperfusion in AMI patients were included. Thirteen RCTs were identified and analyzed. Meta-analysis showed that RIC significantly reduced the area under the curve (AUC) of creatine kinase-myocardial band (CK-MB) (standardized mean difference [SMD] −0.29; 95% confidence intervals [CI] −0.44 to −0.14; P = 0.0002) and AUC of troponin T (SMD −0.22; 95% CI −0.37 to −0.08; P = 0.003). Risk ratio (RR) for ≥70% ST-segment resolution favored RIC group than the control group (RR 1.39; 95% CI 1.03–1.86; P = 0.03). RIC also significantly reduced all-cause mortality (RR 0.33; 95%CI 0.17–0.64; P = 0.001). Subgroup analyses on the CK-MB AUC and ST-segment resolution ≥70% rate showed that the effects of RIC appeared to be affected by the limb used, duration of RIC, and clinical setting. RIC may offer cardioprotective effects by improving ST-segment resolution and reducing the infarct size in AMI patients.

. As shown in Fig. 3A, RIC was associated with a significant reduction in the CK-MB AUC (SMD − 0.29; 95% CI − 0.44 to − 0.14; P = 0.0002) in a fixed-effect model, with no evidence of heterogeneity (I 2 = 0%; P = 0.56). Sensitivity analysis indicated that the omission of anyone trial at each time did not obviously change the pooled SMD and 95% CI. RIC significant reduced the peak CK-MB levels (SMD − 2.37; 95% CI − 3.93 to − 0.81; P = 0.003) in 3 trials 10,15,17 in a random effect model, with evidence of significant heterogeneity (I 2 = 94%; P < 0.001). (Fig. 3B). Two trials 14,15 reported data on peak CK release. As shown in Fig. 3C, RIC was also associated with a significant reduction in peak CK (SMD − 0.38; 95% CI − 0.62 to − 0.13; P = 0.003) compared with control group in a fixed-effect model, with no evidence of heterogeneity (I 2 = 0%; P = 0.49).

Myocardial reperfusion injury as estimated by ST-segment resolution.
Data about RIC on ST-segment resolution ≥ 70% were available in 5 trials 4,8,9,13,14 . As shown in Fig. 5A, the pooled RR for ≥ 70% ST-segment resolution favored RIC group (RR 1.39; 95% CI 1.03-1.86; P = 0.03) than the control group in a random effect model, with evidence of substantial heterogeneity (I 2 = 62%; P = 0.03). In addition, the pooled RR was 1.25 (95% CI 1.09-1.44; P = 0.001) when we changed to a random-effect model. Effect of RIC on ST-segment resolution > 50% was reported in two trials 7,8 . As shown in Fig. 5B, the pooled RR for ≥ 50% ST-segment resolution favored RIC group (RR 1.56; 95% CI 1.18-2.08; P = 0.002) than the control group in a fixed-effect model, with no evidence of heterogeneity (I 2 = 0%; P = 0.42). The pooled RR was 1.51 (95% CI 1.15-1.97; P = 0.003) when we changed to a random effect model.  All-cause mortality. Data about RIC on all-cause mortality were available in 3 trials 8,11,15 . As shown in Fig. 5C, RIC was associated with a significant reduction in all-cause mortality (RR 0.33; 95%CI 0.17-0.64; P = 0.001) in a fixed-effect model during the longest follow-up. There was no evidence of significant heterogeneity (I 2 = 0%; P = 1.00).       Discussion RIC is an easily feasible, well tolerated, and inexpensive technique 19 . A well-designed meta-analysis has evaluated the protective effects of RIC on myocardial injury and clinical outcomes 20 . However, there is high heterogeneity in the studied population, including ST-segment elevation myocardial infarction/urgent PCI, elective PCI, cardiac surgery, congenital heart disease repair, or coronary artery bypass graft. Moreover, this meta-analysis did not particularly address the cardioprotective effects of RIC on the AMI patients undergoing thrombolysis.

Subgroup analyses.
To the best of our knowledge, our meta-analysis specially focused on the cardioprotective effects of RIC induced by intermittent limb ischemia-reperfusion in AMI patients. Our meta-analysis of 13 RCTs involving patients with AMI treated by primary PCI or thrombolysis revealed that RIC induced by intermittent limb ischemia-reperfusion could limit the infarct size as estimated by CK-MB AUC, peak CK-MB release, and troponin T AUC. Moreover, RIC attenuated the myocardial reperfusion injury as estimated by improvement in ST-segment resolution rate.
Troponin was commonly used as a sensitive biomarker for early myocardial injury. In our pooled analysis, RIC significantly reduced troponin T AUC. However, no significant differences were observed between RIC and control group in terms of peak levels of troponin T or troponin I release. These findings may be explained by lack of statistical power due to small sample sizes included in the analysis.
Subgroup analysis showed that on the CK-MB AUC and ST-segment resolution ≥ 70% rate showed that the effects of RIC appeared to be affected by the limb used, duration of RIC, and clinical setting. RIC appeared to have a pronounced effect on the CK-MB AUC in patients undergoing primary PCI than thrombolysis (SMD − 0.46 vs. − 0.23). This finding may be explained by type of cardiac intervention may have different impacts on myocardium, and PCI itself may cause a higher release of cardiac biomarkers. ST-segment resolution has been recognized as a marker of efficient microvascular reperfusion. Resolution of ST-segment deviation after reperfusion is associated with better outcome after ST-segment elevation myocardial infarction 21 . By contrast, rate of ST-segment resolution ≥ 70% was significant in the patients treated with thrombolytic but not in patients undergoing PCI. However, interpretation of our findings should be cautioned due to the small number of trials in the stratified analysis.
This simple intervention is easily applied in AMI patients and may have the potential to reduce cardiac morbidity and mortality. Despite RIC could attenuate cardiac ischemic biomarker release, the effect of RIC on clinical endpoints is conflicting. Our pooled result showed that RIC was associated with a significant 67% reduction in all-cause mortality. However, this finding should be interpreted with caution because the patient numbers were relatively small as well as individual event numbers were low.
There is no standard protocol to induce RIC. Different protocols of RIC may have different cardioprotective effects 22 . RIC stimulus can be applied prior to the intervention, during ischemia, or after blood flow restoration. The timing and site could have potentially affected the cardioprotective effects of RIC. Loukogeorgakis et al. has demonstrated a dose-response protective effect with regard to number of cycles of RIC 23 . In order to achieve the maximal protective effect of RIC, sufficient threshold stimulus should be reached. Our subgroup analyses indicated that the effects of RIC on ST-segment resolution ≥ 70% rate were only statically significant in the RIC duration ≥ 30 min or by the lower limb subgroups. According to these findings, a RIC protocol of at least 3 cycles of 5 min ischemia and 5 min reperfusion (a total duration ≥ 30 min) particularly in the low limb is recommended.
Several potential limitations should be noted. First, this meta-analysis was not based on patient-level data. The potential impact of individual patient data including age, hypertension, diabetes, dyslipidaemia or medications cannot be excluded. Second, infarct size was determined at different time points with a certain degree of clinical heterogeneity. Third, subgroup analysis results were based on the limited number of trials and the small sample size, so these results should be further validated by more well-designed trials. Fourth, apart from all-cause mortality, we did not assess other clinical endpoints because they were only reported in a minority of trials; however, CK-MB or troponin 24 , and ST-segment resolution 25,26 as surrogate indicators can strongly predict clinical prognosis 24 . Fifth, we did not conduct the Begg's and Egger's tests to evaluate publication bias because the included trials were less than the recommended arbitrary minimum number. Finally, this meta-analysis could not determine the optimal protocol of RIC in AMI patients.
In conclusion, RIC induced by intermittent limb ischemia-reperfusion appears to reduce the infarct sizes (determined by AUC CK-MB and troponin T), myocardial reperfusion injury (estimated by ST-segment resolution), and all-cause mortality in AMI patients. However, these conclusions may be not reliable due to insufficient number of trials and the small sample size. More well-designed trials are needed to confirm the cardioprotective effects of RIC in clinical practice.

Methods
Search strategy. The present meta-analysis was performed in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-analysis' (PRISMA) guidelines 27 . The PubMed, Embase, Cochrane Library, VIP, CNKI, and Wanfang database were searched for studies that evaluated the benefits of RIC using intermittent limb ischemia-reperfusion in patients with AMI. The following search terms were used: (RIC OR remote ischemic/ischaemic preconditioing OR remote ischemic/ischaemic perconditioning OR remote ischemic/ ischaemic postconditioning AND myocardial infarction OR AND thrombolysis OR percutaneous coronary intervention OR coronary intervention AND randomized controlled trials OR RCTs. The latest update for literature research was done on November 28, 2016. Additional possible relevant trials were retrieved through a manual search of reference of the included articles.

Study selection.
Trials were considered eligible if they satisfied the following inclusion criteria: (1) RCTs comparing RIC versus no conditioning in patients with AMI; (2) patients were treated by primary PCI or thrombolysis; (3) RIC was induced by intermittent limb ischemia-reperfusion; and (4) trials at least reported one of the following outcome measures, including enzymatic myocardial infarction size as assessed by serum peak creatine Scientific RepoRts | 7:43529 | DOI: 10.1038/srep43529 kinase (CK), peak creatine kinase-myocardial band (CK-MB), CK-MB area under the curve (AUC) as well as troponin I, troponin T or troponin T AUC, electrocardiographic ST-segment resolution (≥ 50% or %70%), and all-cause mortality during the follow-up period. In addition, for the multiple publications from the same population, we chose the article with the complete data. Trials were excluded when: (1) trials consisted of no-AMI patients; (2) trials without reporting any of the outcomes interesting; and (3) non-randomized trials.
Data extraction and quality assessment. Two investigators (CF Man and DD Gong) independently collected data from the included trials. Any disagreements between two reviewers were resolved by consensus. The extracted data included: the first author's surname, year of publication, patients' characteristics, RIC protocol, and outcome measures. For any missing or unclear data, we contacted the correspondence author by e-mail or telephone. The methodological quality of trials was assessed using Cochrane risk of bias tool of RCTs 28 , and grouped as low risk of bias, high risk of bias or unclear risk of bias. Data analysis and synthesis. All analyses were conducted using STATA statistical software version 12.0.
The pooled effect sizes were calculated comparing the RIC to without conditioning, and summarized as a risk ratio (RR) with corresponding 95% confidence interval (CI) for dichotomous data and standardized mean difference (SMD) with 95% CI for continuous data. If continuous data were reported as median ± interquartile range (IQR), the mean and standard deviation (SD) were estimated using the median and the estimator SD = IQR/1.35 28 . Statistical heterogeneity across trials was evaluated using the Cochran's Q test and I 2 statistic. A P-value of Cochran's Q test < 0.10 or I 2 statistic ≥ 50% represented significant heterogeneity. A random-effects model was selected when significant heterogeneity was observed; otherwise, a fixed effect model was used 29 . Subgroup analyses were performed by clinical setting (PCI vs. thrombolysis) and limb used (arm vs. leg). Sensitivity analysis was performed by sequentially deleting anyone study at each turn or replaced by the opposite statistical model to test the reliability of the pooled effect sizes.