Timing of Surgery for Ruptured Arteriovenous Malformations with Intracerebral Hemorrhage: A Systematic Review and Meta-Analysis
Keywords: Arteriovenous Malformation, intracerebral hemorrhage, meta-analysis, systemic review
Arteriovenous malformations (AVMs) are high-flow cerebrovascular lesions having three typical morphological features: feeding arteries, draining veins, and a dysphasic vascular nidus composed of a tangle of abnormal vessels that acts as a shunt from the arterial to venous system. The most common presentation of AVM is intracranial hemorrhage. The AVMs are prone to apoplectic hemorrhage by rupture of flow-related aneurysms, nidal vessels, and draining veins. They are the most frequently detected symptomatic vascular malformations and account for 2% of all strokes and 38% of all intracerebral hemorrhages in patients between 15 and 45 years of age., Right from the diagnosis to multimodality treatment, management of ruptured AVM has undergone drastic evolution over last three decades. But in developing countries like India where there is lack of good transportation facilities, dedicated trained staff, digital subtraction angiography (DSA) lab, and intensive care facilities in a tier-3 city setup, scenario of ruptured AVM management is somewhat different. Microsurgical excision of ruptured AVM with intracerebral hemorrhage (ICH), either early or delayed, is the only strategy possible in such a setup, which provides immediate cure, higher obliteration rate, and the least risk of rebleed. In the literature, strategies of both early surgery and delayed surgery of ruptured AVM have been described with their own merits and demerits.,,,
Although a large numbers of studies have been conducted on early surgery and delayed surgery of ruptured AVM with ICH, no meta-analytical study and large case–control studies have been done so far to reach a definite consensus regarding optimal surgical strategy in patients of acutely ruptured AVM with ICH. So, we have done a systematic review with meta-analysis including our experience to answer the question, “Does early surgery versus delayed surgery for ruptured AVM provide better outcome in terms of morbidity, mortality, and complete excision in patients of ruptured AVM with ICH?”
The above-said PICO (Participant, Intervention, Control, Outcome) question was formulated, and the PICOS approach was defined as shown in [Table 1].
Inclusion criteria for studies
Early surgery- It is defined as surgery for microsurgical resection of AVM within 1 week of ictus.
Delayed surgery- It is defined as surgery performed for microsurgical resection of AVM after 1 week of ictus when brain swelling starts subsiding.
Complete excision- It means no residual AVM in the follow-up angiography (either computed tomography [CT] angiography or DSA).
Good functional outcome- It is the functional status of the patient in the follow-up period, having a modified Rankin score (mRS) of 0–2 or Glasgow outcome score (GOS) of 4–5.
Poor functional outcome- It is defined as the functional status with mRS score 3–6 or GOS 1–3 in the follow-up period.
Types of outcome
Mortality related to ruptured AVM and its management.
Literature search strategy
We conducted a comprehensive online search in June 2020 in PubMed database, Cochrane library, Google scholar, Journal of Neurosurgery, Neurosurgical focus, World Neurosurgery and Child Nervous System journal. In the search process, MeSH terms or key words used in combination were “surgery,” “early surgery,” “delayed surgery,” “elective surgery,” “ruptured,” “intracranial arteriovenous malformations,” and “intracerebral hemorrhage.” The articles available in English language were only screened. The literature search was conducted by two independent reviewers according to the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines as shown in [Figure 1]. Article titles and abstracts were initially screened, and then articles were selected for full-text study. Then, finally, studies fulfilling our inclusion criteria were selected for the meta-analysis.
Data was extracted from the selected studies and recorded in separate Excel sheets (Microsoft Inc., Washington, USA) for early and delayed surgery. The data extracted consisted of name of the author, year of publication, sample size, mean age, clinical condition at the time of admission (Glasgow coma score or World Federation of Neurological surgeons grade), size of ICH, Spetzler–Martin grade of AVM, mortality, number of patients with complete excision of AVM, and number of patients with good functional outcome.
A statistical software package (StatsDirect 2.7.8; StatsDirect, Cheshire, UK) was used to perform all the statistical analyses. We first analyzed the primary and secondary outcomes of early and delayed surgery by calculating the effect size (ES) and 95% confidence intervals (CIs) for each study, and then pooled the data using DerSimonian weights for the random-effects model to derive a pooled ES with 95%CI.
Assessment of heterogeneity
The impact of statistical heterogeneity on the pooled estimates of individual outcomes was assessed using the I2 test, which measures the extent of inconsistency among the results of the studies. An I2 value >40% indicates significant heterogeneity. Heterogeneity was also assessed using the Cochrane Q static, and a P value <0.1 was considered to be significant.
Assessment of publication bias
Publication bias was assessed using the funnel plot and Begg's test., In the funnel plot, X axis depicts the ES and Y axis depicts the standard error of the ES. Symmetrical scattering of ES estimate of studies near summary estimate in the funnel plot shows no publication bias, while asymmetrical scattering is suggestive of publication bias.
Description of studies
Our initial database search retrieved a total of 852 citations, out of which 773 citations were left after removal of duplicates. Of these, 727 citations were excluded on the basis of irrelevant titles and abstracts, leaving 46 articles for full-text assessment. Of these 36 articles were excluded: time of surgery not specified, use of multimodality approach and for various other reasons like results not specified etc. Finally, 10 articles were included in the quantitative synthesis [Table 2] and [Table 3]. Six studies were of early surgery – four retrospective studies and two case series. Three studies were comparative (early vs. delayed surgery), out of which one was prospective and two were retrospective. One study (oral presentation) was a prospective analysis on delayed surgery. Our unpublished retrospective study on early surgery (Varun et al., 2020) was also included to derive combined results on early surgery.
Complete excision rate
The pooled effect by random-effect model showed that delayed surgery of ruptured AVM (ES, 1.00; 95% CI, 0.97-1.00) is slightly better than early surgery of ruptured AVM (ES, 0.96; 95% CI, 0.91 0.99), as shown in [Figure 2]. The ES of early surgery is accompanied by significant heterogeneity (I2 = 44.09%, P = 0.07), while that of delayed surgery is not accompanied by any heterogeneity (I2 = 0.00%, P = 0.46). There is no publication bias on visual examination of funnel plot in early surgery (Begg's test, P = 0.144) and delayed surgery (Begg's test, P = 1.000) groups, as shown in [Figure 3].
The pooled effect demonstrated that delayed surgery (ES, 0.94; 95% CI, 0.86 0.99) is better than early surgery (ES, 0.68; 95% CI, 0.51 0.84) in achieving good functional outcome, as shown in [Figure 4]. The ES of early surgery is accompanied by significant heterogeneity (I2 = 85.63%, P = 0.00), while that of delayed surgery is not accompanied by any heterogeneity (I2 = 0.00%, P = 0.72). There is no publication bias on visual examination of funnel plot in early surgery (Begg's test, P = 0.211) and delayed surgery (Begg's test, P = 0.117) groups, as shown in [Figure 5].
In terms of mortality, the pooled ES demonstrated that delayed surgery of ruptured AVM (ES, 0.00; 95% CI, 0.00 0.01) is slightly better than early surgery of ruptured AVM (ES, 0.04; 95% CI, 0.01 0.10). The ES of early surgery is accompanied by significant heterogeneity (I2 = 53.39%, P = 0.03), while that of delayed surgery is not accompanied by any heterogeneity (I2 = 0.00%, P = 0.96). There is no publication bias on visual examination of funnel plot in early surgery (Begg's test, P = 0.404) and delayed surgery (Begg's test, P = 1.000) groups.
The understanding and management of ruptured AVM have undergone significant improvement with the development of technological advances in imaging modalities, microsurgical techniques, and follow-up outcome data of adjuvant treatment modalities like endovascular and radiosurgical techniques. Various meta-analyses have shown that microsurgical excision is the treatment modality of choice in patients of lower grade AVM (SM grade I–II), which are frequently found in patients of ruptured AVM with ICH., Results of microsurgery in SM grade III AVM are equivocal to radiosurgery and embolization, and in high-grade AVM (SM grade IV–V), the management is still controversial.,
In cases of ruptured AVM with ICH, timing of surgery (early vs. delayed) is still a debatable issue. Early microsurgery in a swollen, acutely injured brain can increase the risk of permanent neurological deficit, and also compression by hematoma may obscure the pure visualization of the true angioarchitecture of the AVM, which may increase the chance of incomplete resection.,, However, several neurosurgeons have rationalized the concept of early surgery on the basis that it quickly reduces the mass effect caused by expansive hematoma, potentially spares the healthy neural tissue from prolonged exposure to toxic blood degradation products, and also prevents secondary deterioration due to peri-hematomal edema.,,,, Although the annual risk of hemorrhage is 2%–4% in unruptured AVM, it is 7%–14% per year for recurrent hemorrhage in the initial 5 years, with the highest risk of re-rupture in the first month of the ictus. With each bleeding episode, the risk of morbidity and mortality increases manifold. Early surgery of ruptured AVM with hematoma evacuation provides complete lesion excision with immediate elimination of the risk of rebleed. In case of delayed surgery, patient must recover twice, first from hemorrhage and later from AVM treatment. Fresh AVM bleeding eases the microsurgical resection of AVM, as the hematoma cavity offers a wide corridor and creates a good cleavage plane, and also, bleeding itself releases a high flow inside the nidus.,, There are very few comparative studies comparing early surgery with delayed surgery for ruptured AVM.,, So, we did a meta-analysis of the results of early and delayed surgery to come to a conclusion.
In our meta-analysis, delayed surgery was found to be better than early surgery in achieving complete excision rate of ruptured AVM. However, there was significant heterogeneity in the results of early surgery group due to difference in patient populations; the heterogeneity could be attributed to the hematoma size or a significant difference in the proportion of AVM grading. Studies of early surgery where evacuation of massive hematoma along with AVM excision was done showed a higher prevalence of residual AVM., This might be due to incomplete visualization of true angioarchitecture of AVM in swollen brain. But on the brighter side, complete excision is achieved in a significant number of patients (96%). Similarly, the overall effect showed that delayed surgery is better than early surgery in terms of good functional outcome and mortality. There was significant heterogeneity in the results of early surgery both for functional outcome and mortality. It may be due to different characteristics of different populations (mean GCS or WFNS score and hematoma size). These results of functional outcome and mortality have significant confounding factor of patient selection bias. Patients with poor clinical condition assigned to early surgery could show poor outcome results, while patients who rebleed and die during waiting for delayed surgery were not considered in delayed surgery results. Also, patients in poor clinical condition after surviving the ictus, who were not considered for delayed surgery afterwards, were also not included in the results. Large comparative studies (early vs. delayed) are needed to do subgroup analysis for age, presurgical clinical condition, and grade of AVM to reduce the selection bias.
Although various confounding factors are present, our meta-analysis supports delayed surgery over early surgery for excision of ruptured AVM. There are several limitations of our meta-analysis. First, omitting non-English studies might lead to publication bias. Second, there was a small number of patients in individual studies. Third, only few comparative studies were available, so subgroup analysis and statistical comparison could not be done. Finally, results of early surgery were limited by the presence of significant statistical heterogeneity, although we used the random-effects model for minimizing the effects of heterogeneity.
Delayed surgery is better than early surgery for excision of ruptured AVM. Larger comparative studies are awaited for subgroup analyses and for consideration of various confounding factors like rebleed, so that more consistent conclusion could be reached. Early surgery for massive hematoma evacuation with AVM excision is the only hope of poor people, as they cannot afford the cost of multiple surgeries and longer hospital stay.
We thank Dr. R. M. Pandey and Dr. Surjeet, Department of Biostatistics, AIIMS, New Delhi for the applied statistics.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
[Figure 1], [Figure 2], [Figure 3], [Figure 4], [Figure 5]
[Table 1], [Table 2], [Table 3]