Efficacy and safety of sonothombolysis versus non-sonothombolysis in patients with acute ischemic stroke: A meta-analysis of randomized controlled trials

Recent studies have shown that inconsistent results of safety and efficacy between sonothombolysis vs. non-sonothombolysis in acute ischemic stroke (AIS). We implemented a meta-analysis to explore the value of sonothrombolysis in AIS treatment. The MEDLINE, EMBASE, and Cochrane Library databases were searched for randomized controlled trials (RCTs) which had evaluated sonothrombolysis or ultrasound thrombolysis in AIS. One hundred five studies were retrieved and analyzed, among them, 7 RCTs were included in the current meta-analysis. In comparison with the non-sonothombolysis, sonothrombolysis significantly improved complete recanalization (RR 2.16, 95% CI 1.51 to 3.08, P < 0.001), complete or partial recanalization (RR 1.90, 95% CI 1.26 to 2.88, P = 0.002), there is also a tendency to improvement of ≥ 4 points in NIHSS score (RR 1.43, 95% CI 0.99 to 2.07, P = 0.057). However, sonothrombolysis and non-sonothrombolysis had insignificant differences in neurological recovery and adverse events. In subgroup analysis, we found that “With t-PA”, “NIHSS > 15”, “Treatment time ≤ 150min”, and “Age ≤ 65 years” are potential favorable factors for efficacy outcomes of sonothombolysis. Sonothrombolysis can significantly increase the rate of recanalization in patients with AIS compared with non-sonothrombolysis, but there is no significant effect on improving neurological functional recovery and avoiding complications.


Introduction
Stroke, along with cardiac disease and cancer are the most common reasons for permanent disability, which is the second leading cause of death worldwide among them [1,2]. Ischemic Outcomes: Efficacy outcomes: excellent functional outcome (modified Rankin Scale (mRS) = 0-1) and good functional outcome (mRS = 0-2) based on mRS, early neurological improvement based on NHISS (NIHSS improve � 4) and recanalization; Safety outcomes: intracranial hemorrhage, death (mRS = 6) and disability (mRS = 3-5). (f) Study years: We searched MED-LINE, EMBASE, and Cochrane Library to find related articles from January 2001 to May 2018. Exclusion criteria were as follows: (a) Study types: case reports, case reviews, retrospective studies, and cohort studies; (b) Control: positive control; (c) Conference abstracts without full text.

Search strategy and information sources
Three major databases: MEDLINE, EMBASE, and Cochrane Library were systematically searched by two authors independently (Z.C. and T.X.). The search strategy of the MEDLINE was to combine all the variables [("ultrasound" AND "thrombolysis") OR "sonothrombolysis"] AND "acute ischemic stroke." Two independent investigators (Z.C. and T.X.) scanned the titles and abstract of all the studies to select applicable studies. The search strategy for EMBASE and the Cochrane Library is similar to what we used for searching MEDLINE. In addition, two investigators (Z.C. and T.X.) manually screened reference lists from RCTs and systematic reviews independently to ensure all relevant studies have been included in this study.

Study selection and data collection
All records from the systematic search in the electronic database and reference lists of RCTs and systematic reviews were evaluated by two authors (Z.C. and T.X.) independently following the eligibility criteria as mentioned above. After strict selection and evaluation, we collect the data from RCTs as follows: basic information on the included trials, inclusion, exclusion criteria for the participants, study design, and outcome assessments ( Table 1).

Risk of bias
The risk of bias plot in individual studies was created using the Review Manager 5.2 software. Uniform criteria of the Cochrane collaboration we applied to assess the risk of bias of RCTs, including selection bias, performance bias, detection bias, attrition bias, reporting bias, and some other potential biases.

Summary measures and synthesis of results
STATA (Version 12.0) was used for assessing the data. Dichotomous outcomes were analyzed as the risk ratio (relative risk [RR]; 95% confidence interval [CI]) and calculated using a random effect model. Statistical heterogeneity was estimated by the I 2 statistic as follows: I 2 < 30% means "low heterogeneity" I 2 = 30 to 50% denotes "moderate heterogeneity," and I 2 > 50% represents "substantial heterogeneity." Subgroup analyses were implemented to detect the application of microbubble and t-PA, the severity of Stroke, time from onset to treatment and patients' ages. Sensitivity analysis was used to explore the stability of the consolidated results. Two-tailed test and a P value less than 0.05 was considered significant for all analyses.

Results
Six hundred thirty-eight titles and abstracts were identified through MEDLINE, EMBASE, and Cochrane Library in total (Fig 1). After removing the duplicates and irrelevant records, 105 articles were included, and 98 articles were excluded because of their types: 23 multiple reports on one RCT, 6 protocol studies, 6 post-hoc analysis, 4 meta-analysis, 5 comments, 32 reviews, and 20 nonrandomized clinical trials. Additionally, 2 conference abstracts without full-text Larrue, V., et al. 2007 [22] and Dinia, L., et al. 2016 [23] were also excluded from the current meta-analysis. Furthermore, the trial of Alexandrov et al. 2008 [24] was excluded due to lose the follow-up information of some patients at three-month (9 out of 15, or 60%) and the data of non-ultrasound control group was the same as CLOTBUST 2004 [17]. Ultimately, seven RCTs [9,[13][14][15][16][17][18] were eligible and contained 549 patients were included in qualitative synthesis (Fig 1). The main characteristics of those included studies are listed in Table 1. All 7 RCTs [9,[13][14][15][16][17][18] include 549 patients are available for the analysis of efficacy and safety outcome.

Subgroup analysis
We performed subgroup analyses to assess the utility of microbubbles or t-PA combined with TCD/TCCS, the severity of stroke at the beginning, time of initial treatment and age of patients.
2. With/without t-PA. The subgroup without t-PA in sonothrombolysis group had a more effective NIHSS improvement (RR 4.66, 95% CI 1.16 to 18.78, P = 0.030; Table 2) compared with non-sonothrombolysis group. However, the subgroup with t-PA in sonothrombolysis group might be safer compared with non-sonothrombolysis group by showing a trend towards the attenuation of death or disability (RR 0.84, 95% CI 0.68 to 1.03, P = 0.091; Table 2).
The sensitivity analysis illustrated that all the consolidated statistics are stabilized.

Risk of bias in included studies
Details about the risk of bias of the included studies are shown in Fig 4. For the random sequence generation assessment, the risk of bias was unclear in 3 out of 7 studies. For the allocation concealment assessment, the risk of bias was unclear in 5 out of 7 studies. For the blinding of participants and personnel assessment, the risk of bias of four trial was unclear and 1 out of 7 trial had a high risk of bias. For the blinding of outcomes assessment, the risk of bias was unclear in 2 out of 7 studies. For incomplete outcome data, the risk of bias of 1 out of 7 trial was unclear and 2 out of 7 trials had a high risk of bias. For the selective reporting assessment, the risk of bias was unclear in 1 out of 7 study. There was no high risk, or unclear risk of bias was observed in other items.

Publication bias
The funnel plot was applied to evaluate the publication bias, indicating that there is no evidence for publication bias (Fig 5). In addition, this evidence is confirmed by a formal statistical test (P = 0.967 of Egger's test).

Discussion
Based on the evidence of the current meta-analysis, sonothrombolysis for AIS might be debatable. When it comes to efficacy issues, we found that sonothrombolysis have a distinct benefit on the recanalization of occlusion, increasing the number of complete recanalization of occlusion, increasing complete or partial recanalization, and decreasing the frequency of the no recanalization than non-sonothrombolysis, which was in accordance with most other studies [5, 9, 14-17, 19, 25]. Although the specific mechanisms of sonothrombolysis were unclear [26], it has been hypothesized that TCD/TCCS could accelerate the movement of fluid around the thrombus so as to enhance the mixture of t-PA into the blood and consequently elevate the concentration of this drug near the occlusion [6,7]. In addition, the pressure waves produced by TCD/TCCS may also augment the permeation of t-PA into the fibrin network and affect the binding of t-PA with fibrin directly [8]. Moreover, some studies believe that sustainable ultrasound exposure may result in vessel vasodilatation probably owing to the elevation of the activity of nitric oxide synthase, which leads to temporal vasodilation which improved reperfusion of local cerebral tissue [27,28].
When it comes to safety issues, sonothrombolysis or non-sonothrombolysis might not influence the occurrence of ICH, disability or death at 3 months in patients with AIS. The results indicated that sonothrombolysis did not increase the adverse events compared with non-sonothrombolysis. In consideration of the sonothrombolysis hypothetical mechanisms, sonothrombolysis may improve the recanalization in short-term, but 3 months' outcomes were impacted by more factors such as age, hyperlipidemia, and recurrence of stroke [6,7,[27][28][29][30]. These factors may lead to the bias of the efficacy and safety results of 3 months' outcomes, so that we may miss the long-term curative effect of sonothrombolysis.
The subgroup analysis on the basis of microbubbles demonstrated that sonothrombolysis without microbubbles was more effective and safer than non-sonothrombolysis thrombolysis through the subgroup data of excellent functional outcome (P = 0.049), good functional outcome (P = 0.032), NIHSS improvement � 4 (P = 0.017) and death or disability (P = 0.011). However, in the sonothrombolysis with microbubbles subgroup, there were no significant differences between sonothrombolysis and non-sonothrombolysis in above mentioned outcome. In theory, the sonothrombolysis plus microbubble therapy should achieve better efficacy than non-sonothrombolysis. However, we got the opposite conclusion according to the sonothrombolysis plus microbubble subgroup anylsis. These results were different from many other articles and trials [9,13,19,31,32] It is possible that only two RCTs nvolve microbubbles, so that a small sample size is difficult to make a difference. In addition, microbubbles were supposed to work by absorbing energy, releasing energy, explosion [12,33] and they might not only accelerate the dissolution of clots [34,35] but also can cause the vascular endothelial damage directly [36]. Therefore, the value of sonothrombolysis plus microbubbles requires more randomized controlled trials to confirm.
Another subgroup analysis, compared with non-sonothrombolysis without t-PA subgroup, sonothrombolysis without t-PA subgroup showed more patients whose NIHSS decrease � 4 (P = 0.030). However, when compared sonothrombolysis with t-PA subgroup with non-sonothrombolysis with t-PA subgroup, there is no significant difference in efficacy outcomes. Taken together, we concluded sonothrombolysis is a valuable method to treat AIS and t-PA also play an important role in thrombolysis. Therefore, t-PA is a favorable factor in sonothrombolysis for AIS.
From the rest of subgroup analysis, we found that sonothrombolysis might be more effective and safer for patients whose NIHSS scores at the beginning of the stroke were higher than 15 or the time of initial treatment was less than 150 min or ages were younger than 65 years, compared with non-sonothrombolysis group.
On the basis of our knowledge, most of previous systematic reviews and meta-analysis brought in several non-randomized types of research [19,25,34,37,38]. Combining all the results of no RCTs was heterogeneous so that these systematic reviews were flawed. Different from above-mentioned systematic reviews, all patients in the present meta-analysis were intervened by sonothrombolysis or non-sonothrombolysis and were randomized, which was the best way to divide risk factors equally over the two groups. This is the second meta-analysis about the sonothrombolysis all evidenced from RCTs (randomized clinical trials). In the first article Ricci et al. 2012 [5] included 5 RCTs, and our meta-analysis included 2 relatively novel RCTs, Dwedar et al. 2014 [16] and NOR-SASS 2017 [18], which was not used in preceding systematic reviews and meta-analysis. These two novel RCTs have a large sample size. Besides, we did further subgroup analysis including 5 relevant factors, which provide more comprehensive comparisons between sonothrombolysis and non-sonothrombolysis. With the data from new RCTs [16,18], we might come to more definitive conclusions in order to guide clinical treatment. The following limitations of our meta-analysis should be noticed. Firstly, this meta-analysis was performed on base on the limited statistics. We only pooled 7 published RCTs 15, 19-24 totally 549 patients were included (sonothrombolysis group, n = 302; control group, n = 247) to analysis the efficacy and safety of sonothromblolysis for AIS. Secondly, the included RCTs showed heterogeneity in the data of complete or partial recanalization (I 2 = 42.2%), excellent functional outcome (I 2 = 44.1%), good functional outcome (I 2 = 60.6%) and death or disability (I 2 = 34.7%). Although the sensitivity analysis demonstrated that all the consolidated statistics were stabilized, these disadvantages of the included studies could not be ignored. Thirdly, there was the possibility of selection bias in our meta-analysis, because we excluded some RCTs only with abstract reported in some meetings like Larrue et al. 2007 [22] and Dinia et al. 2016 [23]. Fourthly, our meta-analysis statistics could not represent all kinds of sonothrombolysis treatment, since we restricted it to TCD/TCCS-induced high-frequency ultrasound treatment (1.8-2 MHz).
In addition to the above limitations, sonothrombolysis itself has limitations in the treatment of AIS. Only patients with occlusion of the proximal middle cerebral artery can benefit significantly by sonothrombolysis. Therefore, the population suitable for treatment is limited. Furthermore, mechanical thrombectomy is currently recommended for the treatment of a large artery occlusion in patients with AIS. Sonothrombolysis is a relatively time consuming and operator-dependent procedure and presents many limitations, mainly technical, according to the presence of a temporal bone window to be performed. However, sonothrombolysis remains a treatment modality when patients admitted to stroke centers without endovascular competence until secondary transport to the intervention center.

Conclusion
In conclusion, this meta-analysis demonstrated that sonothrombolysis treatment for AIS by TCD/TCCSc had significant efficacy on promoting recanalization and the upward tendency of the number of NIHSS improvement � 4. However, it is not significantly different from other efficacy and safety outcomes. Some factors such as with t-PA, NIHSS > 15, treatment � 150 min and Age � 65 years may be potential advantages of sonothombolysis. These evidences can provide confident insights for further research on sonothombolysis in patients with AIS.