What is already known on this topic

Even though it is a well-established procedure, there is no clear consensus in the literature about the efficacy of preoperative embolization of cerebral arteriovenous malformations (AVMs) as adjuvant therapy to surgical resection, adding a non-negligible risk of periprocedural complications.

What this study adds

We assessed baseline cerebral AVM flow with a quantitative imaging technique and compared the intraoperative blood loss (IBL) between patients who underwent combined treatment with preoperative embolization and resection and those who underwent surgical resection only. We also compared baseline AVM flow with post-embolization flow to objectively quantify the impact of embolization treatment prior to surgery.

How this study might affect research, practice or policyIntroduction

The definitive treatment for brain arteriovenous malformations (bAVMs) with high-risk features is usually microsurgical resection for easily accessible lesions or stereotactic radiosurgery for those located in eloquent or deep brain regions.1–3 Embolization has been used for the treatment of bAVMs for almost 60 years, and its current role is mainly adjuvant to microsurgical resection or palliative for non-resectable lesions.1–6

The goal of preoperative embolization is to facilitate surgical resection and to minimize intraoperative blood loss, and the strategy varies according to different AVM anatomic features: elimination of not readily accessible or deeply located feeders, occlusion of flow-related aneurysms or aneurysms distant from the surgical field, and reduction of large active nidus or occlusion of intranidal fistulas, which is usually performed in a number of stages.6–9

There is no clear consensus in the literature regarding the benefit of preoperative embolization on surgical outcome, and the periprocedural risk of morbidity and mortality is not negligible, ranging from 2% to 22% and 0% to 3%, respectively, according to different series.6 10 Studies suggest that preoperative embolization decreases the operative time and the amount of intraoperative blood loss (IBL) with overall better patient outcome.11 12 It has been previously shown that major bleeding during surgery is a significant predictor for permanent neurologic deficit, and IBL could be considered a potential surrogate predictor of patient outcome.13

When planning a treatment strategy for bAVM resection, the risk of periprocedural complications with preoperative embolization should be weighed against the benefit of potential reduction of IBL. However, the impact of preoperative embolization on IBL has shown mixed results in the literature.6 14

In this study we assess baseline bAVM flow with a quantitative imaging technique and bAVM volume prior to any treatment, and we compared the IBL between patients who underwent combined treatment with preoperative embolization and resection and those who underwent surgical resection only. We also compared baseline bAVM flow with post-embolization flow to objectively quantify the impact of embolization treatment prior to surgery.

MethodsPatient selection

Following institutional review board approval, records of patients with supratentorial AVMs who underwent microsurgical resection at our institution between 2012 and 2022 were retrospectively reviewed. Patients who had quantitative magnetic resonance angiography (QMRA) to assess baseline AVM flow before any treatment were included (n=43). Patient consent was not required due to the retrospective nature of the study and de-identification of data.

AVM volume assessment

Volume quantification was obtained from MRI with contrast performed prior to any treatment. Assessment was performed by contouring the AVM nidus on axial, coronal, and sagittal sequences of the MRI brain scan and measuring the maximum diameter in each projection: volume was estimated with the ellipsoid volume equation.15 In the case of ruptured bAVMs, usually the nidus was estimated on MRI performed after near-complete resolution of the brain hematoma or, if in need of emergent resection, it was evaluated after comparison with the diagnostic cerebral angiogram to provide a better estimate of the active nidus in order to minimize confounders due to hemorrhage or mass effect.

AVM blood flow measurement

All patients in the study cohort underwent baseline intracranial QMRA (NOVA VasSol, River Forest, Illinois, USA).16 Follow-up QMRA after the last embolization session was also included, if available. This technique of blood flow quantification in cerebral AVMs has been previously described and validated.17 18

Embolization treatment

Thirty-one of the study patients required preoperative embolization. The treatment strategy was discussed in an institutional multidisciplinary conference with neurosurgeons with both endovascular and open surgical vascular experience. Twenty patients underwent multistage embolization for stepwise flow reduction to minimize the risk of normal perfusion pressure breakthrough. Additionally, high-risk features for intraoperative hemorrhage were addressed (ie, flow-related aneurysm, aneurysm distant from surgical field, deep or not easily accessible feeders). At our institution, n-butyl cyanoacrylate is the embolic agent of choice, and dilution is chosen by the experienced endovascular neurosurgeon according to the target and intranidal flow. In rare cases, for large high-flow AVMs, additional coiling was used to reduce intranidal flow or to create a plug for better control of the glue injection. All embolization procedures were performed under general anesthesia with strict systolic blood pressure control.

Surgical treatment

All patients in the study cohort underwent microsurgical resection under general anesthesia, with strict blood pressure control at induction and during the procedure. After surgical resection was completed, the surgical field was assessed under the microscope with indocyanine-green fluoroscopy to verify completeness of the surgery. IBL was estimated by evaluation of fluid collected in the aspiration canister at the end of the procedure, subtracting the irrigation saline used during surgery, and blood transfused perioperatively. A rough estimate of blood dried with towels and gauzes was added in the computation, with final consensus made between the surgeon and the anesthesiologist.

All patients underwent postoperative diagnostic cerebral angiography within 24 hours of surgery and regular follow-up at 1 year to confirm complete resection, following our institutional protocol.

Statistical analysis

The difference in baseline bAVM flow, bAVM volume, and IBL between the surgery only group and the combined preoperative embolization with surgery group was evaluated with the Wilcoxon–Mann–Whitney test. Linear regression was used to compare IBL between the two groups after controlling for baseline AVM flow and volume. Post-embolization bAVM flow was compared with baseline flow using repeated one-way analysis of variance (ANOVA). All statistical analyses were performed with either Stata/SE (Version 12.0, StataCorp, College Station, Texas, USA) or SPSS (Version 27, IBM Corporation, Armonk, New York, USA).

ResultsCohort characteristics

Forty-three patients met the inclusion criteria with baseline QMRA obtained prior to any treatment. Thirty-one patients required combined treatment with preoperative embolization, 20 of whom had a multistage procedure. Twelve patients underwent treatment with surgical resection only. Nineteen patients had available QMRA after the last embolization session to assess the change in bAVM flow prior to surgery. The characteristics of the study cohort are summarized in tables 1 and 2.

Table 1

Study cohort characteristics

Table 2

Embolization related and surgery related complications

AVM flow, volume, and intraoperative blood loss

Mean baseline bAVM flow (362.3 mL/min vs 89.6 mL/min, p=0.001; figure 1A) and volume (9.6 mL vs 2.8 mL, p=0.001; figure 1B) were significantly higher in the preoperative embolization group than in the surgery only group. There was no significant difference in IBL between the two groups (258.6 mL vs 104.3 mL, p=0.17; figure 1C). Interestingly, once we controlled for baseline AVM flow and volume with linear regression, IBL between the two groups continued to show no significant difference (p=0.53), flow was significantly higher in the preoperative embolization group (p=0.03), and the difference in volume between the two groups was no longer significant (p=0.70).

Figure 1

(A) Comparison of bAVM flow (mL/min) between the surgery only treatment group and the combined treatment group (median 43 mL/min vs 424 mL/min; mean 89.6 mL/min vs 362.3 mL/min; p=0.001). (B) Comparison of bAVM volume (mL) between surgery only treatment group and the combined treatment group (median 1.1 mL vs 6.5 mL; mean 2.8 mL vs 9.6 mL; p=0.001). (C) Comparison of IBL (mL) between the surgery only treatment group and the combined treatment group (median 142.5 mL vs 200 mL; mean 141.3 mL vs 258.6 mL; p=0.17). (D) Comparison of bAVM flow in the preoperative embolization group after the last embolization session and the surgery only group (median 43 mL/min vs 121 mL/min; mean 89.6 mL/min vs 139.5 mL/min; p=0.22). bAVM, brain arteriovenous malformation; IBL, intraoperative blood loss.

Linear regression of IBL controlling for baseline bAVM flow only confirmed the absence of statistical difference between the two treatment groups (p=0.49; figure 2).

Figure 2

Correlation between intraoperative blood loss and baseline brain arteriovenous malformation (bAVM) flow for surgery only group and preoperative embolization treatment group, independently from the initial flow (p=0.49), IBL is <500 mL (except for four outliers).

After embolization, bAVM flow decreased significantly compared with baseline flow (139.5 mL vs 408.0 mL, p<0.001), and the preoperative embolization group showed similar flow to the surgery only group (139.5 mL vs 89.6 mL, p=0.22; figure 1D).

Two outliers for IBL (>1500 mL) were excluded from the study cohort, one in each group.

Embolization-related complications were encountered in one patient who had intraprocedural AVM rupture and bleeding requiring emergent surgical evacuation with residual aphasia due to involvement of the dominant frontal lobe.

Complications related to the surgical procedure occurred in eight patients. Six patients experienced neurologic complications due to the location of the AVM in eloquent areas (three had homolateral hemianopsia, three had mild upper extremity paresis), one patient had a thalamic stroke from posterior cerebral artery thrombosis during surgery, and one had postoperative bleeding in the surgical cavity with worsening of baseline hemianopsia. All complications are shown in table 2.

Discussion

Preoperative embolization as adjuvant treatment to curative resection of bAVM can be used to achieve different goals: to address high-risk features such as flow-related aneurysms at increased risk of rupture, feeders that are deep in the surgical field or not easily accessible, or aneurysms distant from the planned surgical field; or to reduce intranidal blood flow in order to facilitate microsurgical resection. However, the effect of preoperative embolization on surgical performance and overall patient outcome has shown mixed results. Heterogeneity in treatment goals, selection criteria, and unpredictable change in bAVM hemodynamics after partial embolization may account for these uncertain findings.

The morbidity of bAVM preoperative embolization is not negligible, and different series have reported the risk of complications from 8% up to 13%,4 5 7 including a 5.7% risk of hemorrhage, 5.5% risk of ischemia,5 5% risk of permanent neurologic deficit,2 3 risk of recurrence of the unresected embolized portion of the AVM,19 and mortality up to 1%.3 4 In the light of such procedure-related risks, it becomes of paramount importance to define the impact of preoperative embolization on surgical outcome. It has previously been reported that preoperative embolization treatment reduces the risk of surgical resection of large AVMs by making surgical time and intraoperative blood loss comparable to small AVMs,10 11 20 and by showing a similar Glasgow Outcome Scale in the postoperative period.21 However, the AVMs were classified according to the Spetzler–Martin grading system, which provides an estimate of the surgical risk but does not account for AVM flow and nidus characteristics, which are currently easily detectable diagnostic information.16–18

AVM hemodynamics might play an important role in planning the treatment strategy, since it can significantly affect the complexity of surgical resection with a direct impact on IBL, as higher flow bAVMs tend to have higher IBL.6 Wong et al,13 based on the microsurgical cohort of the ARUBA study, showed that major intraoperative bleeding, defined as >1000 mL or >2 units of packed red cells transfusion requirement, was the most significant predictive factor for early and permanent disabling deficit and permanent neurologic deficit, which suggests that IBL might be used as a surrogate measure of high-risk features of bAVMs. Pasqualin et al reported decreased IBL and a reduced risk of postoperative complications with an overall better clinical outcome in patients with bAVMs of similar volume undergoing preoperative embolization compared with patients treated with surgical resection only.12

The effect of preoperative embolization on IBL has, however, shown mixed results in the literature.6 11 12 Donzelli et al 6 found that IBL was not correlated with embolization status but with nidus size and increased AVM flow. In their study cohort, the combined treatment group had larger niduses at baseline, proving that, with embolization, IBL becomes comparable to smaller AVMs. However, in their case series AVM flow assessment was performed with iFlow, a semiquantitative angiographic-based technique whereby the authors selected the most representative feeder to the nidus and its contrast time density curve as a surrogate of AVM flow. The selection of only one feeder for all AVMs and the technique based on contrast density curve over time might represent an oversimplified estimation of the entire AVM flow. In our case series we assessed bAVM flow with QMRA, an objective quantitative imaging technique that has been previously validated.17 18

Mean bAVM initial flow (362.3 mL/min vs 89.6 mL/min, p=0.001) and volume (9.6 mL vs 2.8 mL, p=0.001) in the preoperative embolization group were significantly higher than in the surgery only group. However, IBL was similar in the two groups (258.6 mL vs 104.3 mL, p=0.17), demonstrating that preoperative embolization allows safe resection of complex AVMs with large volume and high flow by achieving IBL comparable to simpler and smaller AVMs. Interestingly, once we controlled for AVM volume and flow with linear regression, initial bAVM flow continued to show a significant difference between the preoperative embolization and surgery only groups (p=0.03) without a significant difference in IBL (258.6 mL vs 104.3 mL, p=0.53). This is probably explained by our institutional practice of performing stepwise flow reduction by exploring the nidus and selectively occluding intranidal fistulas for high-flow bAVMs rather than performing non-specific occlusion of portions of the nidus with the goal of general volume reduction, as is performed in other institutions (see case example in figure 3).

Figure 3

Case example of arteriovenous malformation (AVM) multistage embolization for stepwise flow reduction: baseline MRA Nova map (A), post-embolization #1 (D), post-embolization #2 (G), respective anteroposterior (B,E,H) and lateral (C,F,I) DSA projections. LACA2, left anterior cerebral artery segment 2; LICA, left internal carotid artery; LMCA, left middle cerebral artery; LPCA, left posterior cerebral artery; LVA, left vertebral artery; RACA2, right anterior cerebral artery segment 2; RICA, right internal carotid artery; RMCA, right middle cerebral artery; RPCA, right posterior cerebral artery; RVA, right vertebral artery.

We further provided objective quantification of our embolization strategy by measuring bAVM flow before and after embolization. Repeated ANOVA showed significant flow reduction (from 408.0 mL/min to 139.5 mL/min, p<0.001) with mean residual flow almost comparable to the single treatment group (89.6 mL/min, p=0.22; figure 1D) and with no statistical difference in IBL (p=0.49, figure 2). In accordance with our institution protocol, we generally deem it safe to proceed with microsurgical resection for bAVM with a residual total flow of <200 mL/min and we observed that IBL was <500 mL, except for four outliers (figure 2).

Our complication rate for the embolization and surgical procedure is shown in table 2. One major complication occurred during embolization—namely, AVM rupture that required emergent surgical evacuation after which the patient had residual aphasia. We acknowledge that intraprocedural rupture during embolization is not negligible in our case series, but this retrospective analysis of a small cohort without controlling for AVM size and angioarchitectural risk features does not allow any firm conclusions to be drawn, as the main objective of the study was a quantitative analysis of embolization on AVM treatment. This is a preliminary study that highlights the need for prospective randomized trials to evaluate the risks and benefit of embolization and surgery in the management of bAVM. The major complication related to surgical treatment was a thalamic stroke unrelated to IBL which occurred as a consequence of posterior cerebral artery thrombosis due to surgical manipulation during AVM resection. Even though most of the complications occurred in the surgery and embolization group rather than the surgery only group (7 vs 1), in accordance with our institution protocol patients with AVM who underwent surgical treatment only had lower flow and smaller volume, so they had less complex lesions to resect than the patients in the combined group.

The Spetzler–Martin grading system estimates the surgical risk based on angioarchitectural features, but with recent new insights on AVM hemodynamics, flow assessment provides an additional tool that should be considered in preoperative embolization strategy. In this study we provide objective measurement of bAVM flow and its reduction after embolization treatment, demonstrating the efficacy on surgical outcome, with a significant impact on IBL. This study is the first to provide an objective demonstration of the efficacy of preoperative embolization on reduction of IBL. Due to the lack of consensus on the impact of embolization on surgical outcome and concern for periprocedural morbidity, our study provides further insight on this topic, suggesting the utility of preoperative embolization for safer microsurgical resection of large high-flow AVMs. QMRA provides a non-invasive quantitative measurement of bAVM flow that might help embolization strategy planning.

Limitations

The retrospective nature of the study, the small study cohort, and the incompleteness of the QMRA dataset after the last embolization session are limitations that prevent generalizability of this analysis. The presence of intraparenchymal hematoma introduced bias in baseline flow and volume evaluation. In the majority of cases we made assessment on imaging performed at near resolution of the hematoma, but that was not always possible in case emergent treatment was required.

An additional limitation is the quantification of IBL, which was made as an estimate of fluid collected in the aspiration canister at the end of the procedure, subtracting the irrigation saline used during surgery with the addition of blood transfused in the perioperative period and blood dried with towels and gauzes to the count. Even this is an approximation of actual blood loss; it is based on previous published studies and provides a rough measurement across different patients that can be trended.

The complication rate does not allow any meaningful statistic consideration due to the small size of our cohort. Nevertheless, our main aim was to provide a quantitative assessment of the effect of endovascular embolization on brain AVMs to foster studies on a larger scale to better assess the risk/benefit ratio between the surgical and combined treatment strategies.

Conclusion

An institutional protocol to reduce blood flow selectively and sequentially via embolization in larger and higher flow bAVMs led to IBL comparable to smaller bAVMs undergoing surgical resection only. Therefore, preoperative embolization of large high-flow bAVMs facilitates safer surgical resection by equilibrating IBL.

Data availability statement

Data available on request.

Ethics statementsPatient consent for publication

Not applicable.

Ethics approval

The study obtained Institutional Review Board approval from the University of Illinois College (approval #2016-0659).