Background

Aneurysmal subarachnoid hemorrhage (aSAH) is accounted for 80% of cases of nontraumatic subarachnoid hemorrhage (SAH) [1], contributing to significant mortality. There are two procedures for the treatment of aSAH: microsurgical clipping and endovascular coiling [2]. The first clipping operation was published by Walter Dandy in 1937 [3]. In 1991, the Guglielmi detachable coil for coiling was found, putting a platinum coil into a cerebral aneurysm [4]. The goal of treatment was to occlude the aneurysm to reduce the risk of bleeding. Given this purpose, clipping and coiling are both effective, although there remain controversial with regard to which treatment strategies are better for patients with aSAH.

Until 2002, the International Subarachnoid Aneurysm Trial (ISAT) demonstrated that individuals who underwent coiling were associated with a less morbidity and mortality at 1-year follow-up compared with clipping [5]. This finding contributed to endovascular coiling had been widely accepted becoming the preferred strategy of treatment at many centers [6]. However, the results of ISAT also caused some criticism, such as 7416 of the 9559 patients with ruptured intracranial aneurysms (RIAs) were excluded, the location, and type of intracranial aneurysms (IAs) as well as types of recruiting centers were widely different, and the proficiencies of the performer of coiling and clipping were varied [5, 7].

In recent years, some randomized controlled trials (RCTs) and retrospective comparative studies and prospective studies have also been published, and some results of these publications were different from ISAT [8]. As a result, there remains some debate about the choice of coiling and clipping for patients with aSAH, while it is the aim of this meta-analysis and systematic review to evaluate the two treatments’ effectiveness and complications from a great deal of evidence containing RCTs and observational studies to provide a guiding strategy in selecting which treatment methods to perform in patients with aSAH.

Methods

The Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) [9] was used for this meta-analysis guidelines. And we compared the two treatments by primary outcomes (treatment efficacy) and secondary outcomes (postoperative complications).

Systematic literature search

We searched all literatures with regard to the comparison between coiling and clipping for ruptured intracranial aneurysms (RIAs) through PubMed, MEDLINE, EMBASE, and Cochrane Library databases systematically and comprehensively. The date of these studies was ranged from 2000 to 2021. The search strategies were conducted using “ruptured intracranial aneurysms,” “coiling,” and “clipping,” as our search terms and keywords. A manual search for literatures that was referenced by other publications but met our inclusion criteria was conducted as a supplement. We would use the most current literature, when a study produced multiple papers.

Inclusion and exclusion criteria

Literatures were included if they met the PICOS criteria: (1) population: limited the comparison to the RIAs individuals; (2) intervention: used coiling and clipping; (3) comparison: compared the results after coiling and clipping; (4) outcome measures: the results after treatment and the follow-up; and (5) an official published RCTs or non-RCT.

The exclusion criteria were as follows: (1) Letters to the editor and commentary or conference articles and (2) animal trials; (3) unclear patient outcome data; (4) case reports and case series; (5) systematic reviews or meta-analyses; and (6) other types of IAs, such as trauma.

Selection and data extraction

The data were extracted independently by two observers, C Peng, SF Cai, and YH Diao, containing basic data (author, publication time, age), study characteristics (trial type), and outcomes (rebleeding; mortality; complete occlusion, complications of postoperative) in a table. The senior investigator (YY Yang) would review the data for completeness and accuracy.

Statistical analyses and quality assessment

The results of this study were analyzed by standard software (Stata version 12.0 statistical software). For categorical variable results, risk ration (RR) or odds ratios (ORs) with 95% confidence intervals (CIs) were tested for result assessment. When I2>50%, the data was treated as obvious heterogeneity; therefore, a meta-analysis was performed using random effect model. Otherwise, the fixed effect model was conducted. For continuous variable results, standard mean difference (SMD) or weighted mean difference (WMD) with 95% CIs were calculated for assessment. When I2>50%, the data was treated as obvious heterogeneity, and the data analysis was conducted by a random effect model. Otherwise, the fixed effect model was conducted. The quality of the RCT literatures were assessed by Cochrane Collaboration’s tool, and the Newcastle-Ottawa scale were used to evaluate the quality of the observational studies.

Result

Quality of included studies

The article quality assessment was conducted separately by three reviewers, C Peng, YH Diao, and SF Cai. Thirty-seven observational studies were assessed by the Newcastle-Ottawa scale, and the Cochrane Risk of Bias Tool was used to assess the quality of the 3 RCTs. And the results were showed in Table 1 and Additional file 1.

Table 1 The Newcastle-Ottawa scale for quality assessment observational studies
Full size table

Search results and study characteristics

Initially, 715 literatures were found by searching an electronic database, and 17 articles were identified by manual search. And there were 705 articles after duplicates were removed. 595 publications were deleted by preliminary screening, ultimately, 40 articles met the inclusion criteria and were included in this meta-analysis. The details were shown in the flow chart (Fig. 1). There were 3 RCTs and 37 observational studies [7, 10-43]. A total of 60,875 patients were included and the size of the sample ranged from 32 to 21,905, 31,791 patients were treated by coiling, 29,084 individuals performed by clipping, and other information was shown in Table 2. And synthesis of the results in Table 3.

Fig. 1
figure 1

Forty articles met the inclusion criteria and were included in this meta-analysis

Full size image
Table 2 Overview of Included Studies
Full size table
Table 3 Meta-analysis results
Full size table

Effectiveness of treatment

Modified Rankin Scale (MRS)

The mRS were used in this article to assess the quality of life. There were 11 articles, 4106 patients, including the result of mRS. 65.2% of the patients in the coiling group and 59.9% of patients in the clipping group had a good quality of life defined as mRS0-2. And there was a statistical significance in the results of mRS0-2 (coiling 1523 of 2336 (65.2%) VS clipping1454 of 2428 (59.9%); OR=1.327; CI=1.093–1.612; p<0.05; I2=38.8%; Fig. 2).

Fig. 2
figure 2

Statistical significance in the results of mRS0-2

Full size image

Rebleeding

Fourteen articles included a total of 4659 patients with RIAs provided the rate of rebleeding after clipping or coiling. There was higher postoperative rebleeding in the coiling group than in the clipping group. And it was associated with a statistical significance (coiling 128 of 2232 (5.7%) VS clipping103 of 2427 (4.2%); RR=1.410; CI=1.092–1.822; p<0.05; I2=10.6%; Fig. 3).

Fig. 3
figure 3

Fourteen articles included a total of 4659 patients with RIAs provided the rate of rebleeding after clipping or coiling. There was higher postoperative rebleeding in the coiling group than the clipping group. And it was associated with a statistical significance

Full size image

Mortality

Twenty-one literatures encompassing the rate of mortality after coiling or clipping among 44,909 patients with RIAs. Coiling had a significant effect on the risk of mortality compared with clipping (coiling 3847 of 25,268 (15.2%) VS clipping 2955 of 19,641 (15.0%); OR=1.116; CI=1.054–1.180; p<0.05; I2=36.9%; Fig. 4).

Fig. 4
figure 4

Twenty-one literatures encompassing the rate of mortality after coiling or clipping among 44,909 patients with RIAs

Full size image

Complete occlusion

Eight studies included the result of complete occlusion, and the result was high heterogeneity. This study deleted a publication by heterogeneity analysis (Fig. 5). Seven studies included the result of complete occlusion among 2545 patients with RIAs. There was a higher rate of occlusion in the clipping group than the coiling group with a statistical significance (coiling 956 of 1480 (64.6%) VS clipping 881 of 1065 (82.7%); OR=0.375; CI=0.308–0.456; p<0.05; I2=0.0%; Fig. 6).

Fig. 5
figure 5

Eight studies included the result of complete occlusion, the result was high heterogeneity

Full size image
Fig. 6
figure 6

Eight studies included the result of complete occlusion, the result was high heterogeneity

Full size image

Postoperative complications

Vasospasm

Thirteen publications included a total of 2857 patients with RIAs who provided the result of vasospasm after clipping or coiling. There was a less postoperative vasospasm in the coiling group than in the clipping group with a statistical significance (coiling 241 of 1177 (20.5%) VS clipping 416 of 1680 (24.8%); OR=0.787; CI=0.649–0.954; p<0.05; I2=41.1%; Fig. 7).

Fig. 7
figure 7

Thirteen publications included a total of 2857 patients with RIAs provided the result of vasospasm after clipping or coiling

Full size image

Hydrocephalous

Nine literatures contained the result of hydrocephalous after treatment among 3856 patients with RIAs. Coiling had a significant effect on the postoperative hydrocephalous compared with clipping (coiling 611 of 1819 (50.6%) VS clipping 581 of 2037 (39.9%); RR=1.143; CI=1.043–1.252; p<0.05; I2=30.7%; Fig. 8).

Fig. 8
figure 8

Nine literatures contained the result of hydrocephalous after treatment among 3856 patients with RIAs

Full size image

Cerebral infarction

There sixteen articles concluded the result of ischemic infarct after coiling or clipping among 5423 patients. Coiling had a lower postoperative ischemic infarct than clipping with a statistical significance (coiling 375 of 2598 (14.4%) VS clipping 597 of 2825 (21.1%); RR=0.669; CI=0.596–0.751; p<0.05; I2=18.9%; Fig. 9).

Fig. 9
figure 9

There sixteen articles concluded the result of ischemic infarct after coiling or clipping among 5423 patients

Full size image

Postoperative neuro deficits

The five articles concluded the result of neuro complications (defined as any new weakness, decreased level of consciousness, paresthesia, or cranial nerve deficit), after coiling or clipping among 3076 patients. Clipping had a higher rate of postoperative neuro deficits than coiling with a statistical significance (coiling 119 of 1530 (7.8%) VS clipping 167 of 1546 (10.8%); RR=0.720; CI=0.582–0.892; p<0.05; I2=15.6%; Fig. 10).

Fig. 10
figure 10

There five articles concluded the result of neuro complications (defined as any new weakness, decreased level of consciousness, paresthesia or cranial nerve deficit), after coiling or clipping among 3076 patients

Full size image

Discussion

This meta-analysis summarized the available data with regard to outcomes of patients with RIAs who underwent clipping or coiling procedures systematically. Our meta-analysis included 40 articles involving 60,875 patients with RIAs. And we compared eight outcomes between coiling and clipping including the effectiveness of treatment (mRS, postoperative rebleeding, postoperative mortality, the rate of complete occlusion) and the postoperative complications (vasospasm, hydrocephalous, cerebral infarction, postoperative neuro deficits).

This meta-analysis showed that patients who underwent coiling had a significantly better quality of life (mRS 0-2) than those who underwent clipping at 1 year after treatment. Liu et al. [36] also reported that coiling patients had more good quality of life outcomes than clipping patients at 1 year after treatment. And this result was consistent with ISAT data [5, 7]. Additionally, some articles [23, 28] showed the trend that coiling was related to a higher rate of good outcomes (mRS 0-2) than the clipping group. Yu et al. [21] reported that the result of Glasgow Outcome Score (GOS) (1–3) was lower in endovascular coiling (12/80, 15%) than in microsurgical clipping (30/89, 34%; p<0.05). Zhang et al. [19] had the opposite result about the rate of GOS (4–5). Because the admission grade (Hunt-Hess 4–5; p<0.01) [27] was associated with poor outcomes, it could explain why there were different results.

250 (23.5%) of 1063 individuals who underwent coiling treatment were dependent or dead at 1 year, compared with 326 (30.9%) of 1055 patients with clipping, an absolute risk reduction of 7.4% (95% CI 3.6–11.2, p<0.05) reported by Molyneux et al. [7]. Spetzler et al. [43] also showed coiling was related with a lower rate of mortality. While Shen et al. [18] had the opposite point, their result showed coiling was associated with a higher mortality rate than clipping, this result was similar with our meta-analysis. Our result of mortality was different from published studies, and the difference in categorical data may be one of the reasons [18]. Additionally, this study found that the coiling group has a higher incidence rate of rebleeding rate and a lower complete occlusion rate. It may be associated with higher mortality in the coiling group.

Several articles [18, 19, 31, 32] demonstrated that a trend toward postoperative rebleeding in the coiling group, while other literatures [17, 33] showed the clipping group had a higher rate of rebleeding than the coiling group, and there was no significant difference in their results. In the present article, we find a significantly higher risk of rebleeding in the endovascular coiling group (p<0.05). Varelas et al. [33] reported that rebleeding was significantly associated with the ventriculoperitoneal shunt (p<0.05), and some published articles suggested that rebleeding also depended on the follow-up period and on the rate of occlusion after endovascular coiling or microsurgical clipping [5, 7] and this meta-analysis also found that clipping was significantly associated with a higher rate of complete occlusion (p<0.05), this result was consistent with published studies [28, 31, 35]. Murayama et al. [44] also reported that the rate of complete occlusion was found in 55% of aneurysms, and the lesion neck remnant was identified in 35.4% of aneurysms and the rate of recanalization was up to 20.9%, which was associated with the neck of the aneurysm and size of the dome. And coil compaction and/or loosening and a high rate of the remnant of the neck could also cause recanalization [5, 45].

Our articles showed endovascular coiling was associated with a significantly lower risk of vasospasm, cerebral infarction, post neuro deficits, but with a significantly higher postoperative hydrocephalous than microsurgical clipping.

Li et al. [35, 46] also showed the lower incidence of vasospasm and cerebral infarction in the coiling group. Some other publications [46, 47] were similar to ours about the infarction. One of the vasospasm reasons is that blood degradation products, accumulating in subarachnoid space and reserve as triggers to cause intramural inflammation and endothelial dysfunction [48]. However, there was an argument about vasospasm, someone thought that remove the cisternal blood during clipping would reduce the risk of vasospasm [45]. But this effect could be offset by other effects related with clipping [49], such as surgical operations of the vessels and craniotomy with brain retraction would aggravate the preexisting cerebral vasospasm. And some previous publications suggested that cerebral vasospasm was associated with the incidence of cerebral infarction [50, 51]. There were some other reasons of cerebral infarction: microsurgical clipping blocked some microvascular during surgery, leading to ischemia event. The compression of the small vessels that around the lesion clip may lead to local ischemia [18]. These factors may cause a higher risk of infarction in the clipping group. Additionally, vasospasm-related cerebral infarction significantly influences the rate of mortality following aSAH and cause poor clinical outcomes [52].

The result of postoperative neuro complications was consistent with some published studies [49, 53], and Dumont et al. also analyzed the risk factor of neuro deficits, such as clipping, ventriculostomy, thick clot size, history of hypertension, and intracerebral hemorrhage [49].

So far, some publications had reviewed the morbidity of hydrocephalus after endovascular coiling and microsurgical clipping systematically, while there was no uniform conclusion [21, 54]. While the result of Shen et al. [18] was consistent with this article that coiling was related with a higher risk of hydrocephalous. As is known to all, arachnoid granules absorbed cerebrospinal fluid (CSF), and some CSF was absorbed through the cerebral capillaries. Blood clots may lead to impairment of CSF absorption by disturbing cerebral capillaries and arachnoid villi, causing cerebral hydrocephalous [54]. While clipping could remove the blood clots, improving circulation of CSF, decreasing the risk of hydrocephalous [18]. And the controversy with regard to the result of hydrocephalous may be the different diagnosis criteria of cerebral hydrocephalus [19].

This study has several potential limitations: (1) The included literatures were only 3 RCTs, and this article was limited to the evaluation of short-term results. (2) The sample of some comparative indicators was relatively small.

Conclusion

Coiling was significantly associated with a better quality of life (mRS0-2), a lower incidence of postoperative complications (vasospasm, cerebral infarction, neuro deficits), and a higher rate of mortality, rebleeding and hydrocephalous than clipping. Additionally, coiling was associated with a lower rate of complete occlusion.