Introduction

Arteriovenous malformations (AVMs) constitute a specific subgroup of lesions that can affect each structure of the brain, with a prevalence between 70 and 93% at the level of supratentorial structures.1 They consist of an anomalous tangle of modified, high-flow vessels characterized by the shunting of arterial blood into draining veins in the absence of a physiological capillary bed. The average reported detection rate for this condition is 1.34 per 100,000 person-years.2 Although AVMs can be silent for many years, in more than 50% of the cases clinical onset occurs with hemorrhage.2 About 2% of all intracranial hemorrhage is due to rupture of a brain arteriovenous malformation (bAVM). AVM-related hemorrhage results in varying degrees in morbidity with a risk of permanent neurological deficits of 30 to 50% and mortality of 10 to 30%.3 Current literature data show that AVM rupture is associated with a risk of rebleeding ranging from 6 to 39% per year following the initial hemorrhage.4 5 6

While definitive treatment is fairly standardized,7 surgical timing in case of ruptured bAVM is less standardized. International literature2 8 9 10 recommends, in general, a variable “rest period” (1–6 weeks) between the hemorrhage and the conclusive treatment.

The aim of the study was to clarify, through a retrospective analysis of a multioperator series, (1) which parameters should be considered in the choice of the best surgical timing in case of ruptured bAVMs and (2) how these parameters can influence outcomes.

Materials and Methods

Between January 2010 and June 2018, 32 patients affected by bAVMs underwent surgery in the Neurosurgery Department of Sapienza University of Rome. Of these, 25 met the inclusion criteria (radiological evidence of intracranial hemorrhage) and were therefore analyzed.

For each patient we evaluated: age, sex, pre-existing medical comorbidities, onset symptoms and signs, discharge symptoms and signs, angioarchitectural characteristics of the lesion (including size, location, and venous drainage), bleeding severity, treatment modalities, surgical timing, and outcomes at 3 and 12 months. AVMs were investigated using computed tomography angiography, magnetic resonance imaging angiography, and/or traditional angiography. The Spetzler-Martin AVM grading system was applied to the classification of the lesions.11 The presence or absence of aneurysms on feeding arteries was also taken into account. Risk stratification, in relation to bleeding severity, was evaluated through the use of intracerebral hemorrhage (ICH) score that takes into consideration: (1) Glasgow Coma Scale (GCS), (2) hematoma volume (mL), (3) infratentorial origin (yes/no), (3) intraventricular hemorrhage (IVH) (yes/no), and (4) age (years). Regarding surgical timing, we divided patients into two groups: (1) “early surgery” (timing ≤ 48 hours, 11 patients) and (2) “delayed surgery” (timing> 48 hours, 14 patients). In each case, histological examination confirmed the diagnosis of AVM.

ICH score and time of surgery for each patient are showed in Table 1.

We employed the modified Rankin Scale (mRS) to assess outcomes in terms of degree of disability/dependence following the acute event.12

Univariate and multivariate analysis of the collected data were performed using Excel, MATLAB and Stata. For each parameter, we calculated mean and standard deviation. We also estimated correlation, covariance, and relation coefficient with the ordinary least squares (OLS) linear regression model.

Results

Results of the cohort analysis are described in Tables 1 2 3 4, and in Figs. 1 2.

Close

Fig. 1 Clinical presentations.

Export to PPT

Fig. 1 Clinical presentations.

Close

Fig. 2 Arteriovenous malformation (AVM) location.

Export to PPT

Fig. 2 Arteriovenous malformation (AVM) location.

Trying to respond to our objectives, we focused our attention on the parameters analyzed below.

ICH Score, Surgical Timing, and Outcome

The mean ICH score for the “early surgery” subgroup of patients was 2.82 ± 0.71, while it was 1.14 ± 0.81 for the “delayed surgery” one. Patients with a favorable outcome (mRS≤2) at 3 months and 1 year had a mean ICH score, respectively, of 1.26 ± 0.72 and of 1.57 ± 0.92, while those with a poor outcome (mRS>2) at 3 months and 1 year had a mean ICH score respectively of 2.8 ± 0.71 and of 2.8 ± 0.63. We found a positive statistical correlation (ρ) between bleeding severity and outcome, given the calculated values: 0.63 at 3 months and 0.50 at 1 year. Covariance (σ) (ICH-score, mRS) at 3 months and 1 year was respectively 1.24 and 1.07, both positive (the two values positively covary together).

Besides, through a linear regression model (OLS), we calculated the regression coefficient of the variable ICH score against mRS, obtaining these results: 0.91 at 3 months and 0.79 at 1 year. So, an increase in 1 unit of the ICH score implies an increase in the expected mRS score of 0.91 at 3 months and of 0.79 at 1 year (Figs. 3 4).

Close

Fig. 3 Effect of intracerebral hemorrhage (ICH) score on modified Rankin Scale (mRS). Regression coefficient at 3 months.

Export to PPT

Fig. 3 Effect of intracerebral hemorrhage (ICH) score on modified Rankin Scale (mRS). Regression coefficient at 3 months.

Close

Fig. 4 Effect of intracerebral hemorrhage (ICH) score on modified Rankin Scale (mRS)s. Regression coefficient at 1 year.

Export to PPT

Fig. 4 Effect of intracerebral hemorrhage (ICH) score on modified Rankin Scale (mRS)s. Regression coefficient at 1 year.

Hematoma Volume, Surgical Timing, and Outcome

In the “early surgery” subgroup, the average hematoma volume was 45.55 ± 23.21 mL, while in the “delayed surgery” subgroup was 29.89 ± 21.33 mL. As for the relationship between hematoma volume and outcome, patients with a volume <30 mL (standard cut-off volume used for the evaluation of ICH severity13) had an average mRS of 2.09 ± 1.13 at 3 months and of 1.54 ± 1.01 at 1 year. In contrast, patients with a volume >30 mL had an average mRS of 3.23 ± 1.31 at 3 months and of 2.38 ± 1.91 at 1 year.

We also calculated the regression coefficient of variables from here on (i.e., volume, age, GCS, infratentorial hemorrhage, IVH) in relation to outcome. We excluded ICH score, being this a synthesis of the other variables.

The following linear model was applied:

Regarding volume, we obtained the following coefficients: 0.02 at 3 months and 0.016 at 1 year, indicating that an increase in 1 mL in the hematoma volume relates to an “expected” increase of 0.020 at 3 months and of 0.016 at 1 year in terms of mRS.

Discussion

Current neurosurgical literature still lacks randomized controlled trials14 showing that the benefits of the definitive treatment of bAVMs outweigh the risks. However, given that hemorrhage at diagnosis is one of the most relevant risk factors for subsequent new hemorrhage,3 definitive treatment is widely recommended,7 14 15 16 17 with the exception of those patients for which, because of the high surgical risk of the lesion (Spetzler-Ponce class C or Spetzler-Martin grades IV and V), conservative management is recommended.16 17 18

The outcome of patients with ruptured bAVMs is generally favorable, with a mortality rate and a permanent disability rate relatively low compared with those of spontaneous intracerebral hemorrhage (sICH).

International literature2 7 8 9 10 recommends, in general, a “rest period” (1–6 weeks) between the hemorrhage and the conclusive treatment of the disease, since the risk of early rebleeding, though possible, is rare, and the “parenchyma” of the lesion can be “friable” in the period following the hemorrhage thus increasing surgical risk. It has been stated that this approach allows the safest surgical treatment (when this is indicated), also through a better diagnostic evaluation of the problem. However, when (1) significant mass effect is present, or (2) it is not possible to stabilize the patient and conservatively control intracranial and cerebral perfusion pressures, with subsequent progressive neurological deterioration, an urgent evacuation of the hematoma may be necessary.10 19 Management of this situation may also include the early resection of the AVM in the same single surgical time. AVM definitive treatment should be considered for those injures that are not expected to have a high surgical risk8; in this sense, there are also studies showing that this management is associated with a very favorable outcome.9

In our study, we found that, in case of hemorrhagic bAVMs, ICH score evaluation may be decisive in the choice of the best surgical timing. ICH score can also be one main predictive factor for patient outcome (in particular for early outcome), in accordance with Hemphill et al.20 This finding suggests that ICH score, scarcely used in presence of hemorrhagic bAVMs, can be a useful tool for both decision-making regarding therapeutic management, and outcome prediction, especially when considered along with two of the other characteristics examined, that is, age and Spetzler-Martin grade.21 The ICH score parameters that seemed to influence surgical timing the most were GCS and hematoma volume.

As far as outcomes are concerned, regression coefficients allowed us to detect a significant correlation with GCS, hematoma volume, and infratentorial origin.

It has been extremely interesting to note that while GCS and hematoma volume seem to affect especially early outcomes, the presence of an infratentorial hemorrhage affects both early and late outcomes. Infratentorial origin of the hemorrhage should, therefore, be considered in the decision-making about surgical timing. Moreover, also sICH guidelines emphasize that, in the presence of a cerebellar hemorrhage determining neurological deficits or hydrocephalus from IV ventricle obstruction, the early evacuation should be considered.22

Age deserves a separate discussion, as it is a variable that strongly influences outcomes.

Patients under 40 years of age had a significantly better outcome than patients older than 40 years. This parameter may explain why the outcome of patients affected by intracranial hemorrhage from ruptured bAVM is usually better than that of patients affected by primary sICH, being these patients generally older. Accordingly, in a population-based study released in 2008, van Beijnum et al found that in case of AVM-related ICH, patients were younger, had a lower pre-stroke blood pressure, and a higher GCS at the entrance. They also were more likely to have an atypical ICH (lobar location) compared with patients affected by primary sICH.23

We have already argued about surgical risk determined by AVM characteristics.11 18 24

Certainly, also in our case, the grade of AVM influenced surgical timing. Nearly everyone within the “early surgery” group harbored a grade I or II AVM (Spetzler-Ponce class A), only one patient had a grade III and one patient a grade IV. Instead, 50% of patients in the “delayed surgery” group presented a grade III AVM (Spetzler-Ponce class B). Grade III AVMs are injuries for which current literature recommends a careful angiographic assessment and multimodal treatment, wherever possible.18

In our study, early outcomes seem more influenced by the ICH score and therefore from hemorrhage severity, while delayed outcomes by Spetzler-Martin grading; this confirms the role of AVM angioarchitectural anatomy in the preoperative surgical risk assessment with the purpose to guide surgeons in the choice of the best treatment possible.16 17 18 It is, so, actually preferable (when possible) to intervene on the malformation away from the hemorrhage, to assess the best therapeutic strategy.16 17 According to Lawton et al,17 if a patient would have an additional benefit from embolization (provided that the risk is tolerable), double modality management appears indicated; of course, if the risk associated with embolization seems too high (e.g., in case of elevated intracranial pressure) only microsurgery should be done. Our analysis also showed that patients under 40 years of age with poor neurological status at the entrance, average ICH score of 2.67 ± 0.62, and low AVM surgical risk, undergoing early resection of the lesion, had a 3-month poor average mRS (2.47 ± 0.71), while at 1 year the outcome was very favorable (1 ± 0.4). This suggests, in line with other studies,9 18 25 how in young patients with a high ICH score at the entrance, one or more risk factors consistently related to poor outcome (low GCS, volume clot > 30 mL, infratentorial hemorrhage), and a low-grade Spetzler-Martin malformation, early intervention of evacuation and AVM removal in one single surgical time represents a valid and feasible therapeutic strategy. In a study released in 2020, Deng et al reported a single-center study on 111 hemorrhagic bAVMs in children suggesting that early surgical treatment is preferable following the hemorrhagic presentation.25 Aboukaïs et al26 found that in grade I bAVMs long-term morbidity is, in the majority of cases, related to the brain damage determined by the hemorrhagic event; only in few cases, this may be linked to AVM resection.

Main limitations of this study are represented by the small cohort dimension, and by the retrospective fashion of the study.

Conclusions

When used together with Spetzler-Martin grading scale, ICH score seems to be a valid clinical instrument both in the management decision-making and in the outcome prediction in patients affected by hemorrhagic bAVMs. These parameters, considered together, allow a better evaluation about surgical timing, kind of treatment, and probable outcome.

Early outcomes seem to be more influenced by ICH score, and therefore by bleeding severity, delayed ones more by Spetzler-Martin grade, and so by AVM angioarchitecture. This suggests that it is indeed preferable, when possible, to intervene on the lesion itself after some time from the hemorrhage to evaluate the best treatment strategy.

Our study contributes to affirm the importance of an early treatment in patients affected by grade I or II bAVMs though with poor neurological status at the entrance. In fact, in young patients with a high ICH score at the entrance, one or more risk factors consistently related to poor outcome (low GCS, volume clot > 30 mL, infratentorial hemorrhage), and a low-grade Spetzler-Martin malformation, early intervention of evacuation, and AVM removal in one single surgical time represent a valid and feasible therapeutic strategy. Confirmation of this will, of course, require further study and larger patient cohorts.