Demographic and Angioarchitectural features associated with Brain Arteriovenous Malformations Hemorrhagic Presentation , in South Africa

Background: Intracerebral hemorrhage (ICH) represents 10% of all strokes and 2% of spontaneous ICH are due to ruptured Brain Arteriovenous Malformations (BAVMs). The morbidity and mortality associated with AVM hemorrhage is as high as 35 and 10%, respectively. This study aimed to investigate the association between specific angioarchitectural features, demographic variables and hemorrhagic presentation of BAVMs. Methods: We conducted a retrospective observational chart review of all consecutive patients who presented with brain AVMs in KwaZulu-Natal, South Africa, over a period of 10 years. The demographic and clinical presentations were derived from patient’s electronic medical record; Soarian® [Siemens Healthcare GmbH, Erlanger, Bavaria, Germany]. Radiological features were determined using axial Computerized Topography (CT) or Magnetic Resonance Imaging (MRI) Scan. Angioarchitectural features were determined from Digital Subtraction Angiography (DSA). Simple and multiple logistic regression models were used to identify factors associated with brain AVM hemorrhage. Results: This study identified 157 patients with brain AVMs. Seventy-one (45%) patients presented with hemorrhage. Statistical analysis found deep AVM location (OR = 3.3; 95%CI: 1.4−7.7), exclusive deep venous drainage (OR = 3.3; 95%CI: 1.1–9.9) and presence of aneurysms (OR = 3.4; 95%CI: 1.4–7.9) to be significant predictors of brain AVM hemorrhage. Conclusion: The significant morbidity and mortality associated with brain AVMs hemorrhage behooves us to identify features that predispose these lesions to bleed. These angioarchtectural features may serve as therapeutic targets to reduce the risks of brain AVM hemorrhage.


Introduction
Arteriovenous malformations of the brain are associated with one of the most devastating complications affecting the central nervous system, one of which being hemorrhagic stroke [1][2][3][4].Several prospective population based studies suggest that more than half of all brain AVM patients may suffer intracranial hemorrhage [5][6][7][8][9].About 2% of all spontaneous intracerebral hemorrhages are due to ruptured brain AVMs [10][11][12], and mortality and morbidity associated with each hemorrhagic episode is estimated to be 10% and 30-50%, respectively [13,14].If left untreated, recurrent hemorrhage is estimated between 12-18% [10] in the first year following initial hemorrhage and fall to an annual rupture rate of 6-15% with devastating consequences [14].
The highest incidence of recurrent hemorrhage happens within the first year following the initial event [15,16].Choi, et al. [10] and Al Shahi, et al. [16] found the re-bleed rate within the first year following initial hemorrhage to be 12 and 18%, respectively.Brown, et al. [15] found that actuarial risk of hemorrhage in unruptured brain AVMs was 1.3% per year in the first year, 1.7% per year at 5 years, 1.5% per year at 10 years and 2.2% per year at 15 years.The estimated cumulative lifetime risk of hemorrhage was found to be around 29% [17].The mortality associated with brain AVM hemorrhage was reported to be 29% over a follow-up period of 8.2 years [15].Treatment should therefore aim for early therapy in those patients who present with ruptured brain AVMs.The risk of severe morbidity amongst the hemorrhage survivors is estimated to be around 23% in some population based studies [10,15].Crawford, et al. [18] reported overall annual mortality rate of 1.5% per year.They reported 42% risk of hemorrhage, 29% risk of death, 27% risk of neurological deficit by 20 years follow-up of conservatively managed brain AVMs.
Although brain AVMs represent only 2% of all intracranial hemorrhage [10], they typically affect otherwise healthy young adults between age group 20-40 years [8,19], inflicting significant morbidity upon the highly economically productive members of the society, with resultant heavy social burden on the society and health system.Therefore, an understanding of the factors that predict risk of hemorrhagic presentation is needed to formulate therapeutic strategies in order to eradicate or minimize the risk of hemorrhage.
We undertook the study to look at the association between certain angioarchitectural features and demographic variables, with initial presentation of hemorrhage in our institutions.To our knowledge, no such study has been published in Africa.

Methodology Study Location and Design
A retrospective observational comparative chart review conducted in accordance with Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines was performed on all consecutive patients who presented to three interventional neuroradiology hospitals with brain AVMs in the Province of KwaZulu-Natal, South Africa, over period of 10 years.

Data Collection
Data was collected from password secured electronic database (Soarian ® ) [Siemens Healthcare GmbH, Erlanger, Bavaria, Germany] and Syngo.plaza ® [Siemens Healthcare GmbH, Germany] on 158 patients who were diagnosed with brain AVMs from May 2005 until May 2015.The demographic variables collected for analysis included; age, race and gender.The mode of clinical presentation included the presence or absence of intracranial hemorrhage at initial presentation.The demographic and clinical presentation information was derived from patient's electronic medical records, Soarian  and clinical charts.Radiological and angioarchitectural features derived from Syngo.plaza ® included; brain AVMs location, nidus size, venous presence and ISSN: 2373-8995 the type of associated aneurysms and the venous drainage patterns.One patient was excluded from the study due to missing data in all important variables thus leaving 157 eligible patients in the study.
Brain AVMs size and locations were determined using Computerized Topography (CT) scan and the Magnetic Resonance Imaging (MRI).Sizes were categorized according to the Spetzler-Martin scale method into <3 cm, 3-6 cm or >6 cm, and further sub-grouped into Spetzler-Ponce grades 1, 2 and 3. AVM location was categorized into deep and cortical-lobar.Deep location was defined as larger portion of the nidus located deep within the white mater tracts, basal ganglia, thalamus, periventricular region or posterior fossa.Angioarchitectural features were determined from the digital subtraction angiography (DSA).The nomenclatures for the variables were defined according to the Joint Writing Group of the Technology Assessment Committee where applicable [20].All brain AVMs angiograms and radiological images were evaluated by a senior interventional neuroradiologist with more than 20 years of experience in order to limit inter-observer variability.

Statistical Analysis
Data analysis was done using STATA (version 14) [StataCorp LLC, College Station, Texas, United State of America].The demographic, clinical and angioarchitectural features were reported using medians (interquartile ranges) or means (standard deviations) for continuous variables and frequencies (percentages) for categorical variables.For continuous variables, the Wilcoxon rank sum test was used to compare medians.The degree of association between categorical variables was assessed using Pearson chi-square test or Fisher's exact test where appropriate.A p -value of less than 0.05 was considered to indicate statistical significance.Simple logistic regression analysis was performed to identify factors associated with hemorrhagic presentation.Factors with a p -value <0.2 in the univariate analysis, except for demographic factors, were entered into multiple logistic regression analysis to determine predictors of brain AVM hemorrhagic presentation.

Ethical Consideration
The biomedical research ethics committee (BREC) of the University and the provincial health research committee, granted approval for this study (BREC no: BE 497/15).

Patient's Consent
This study is a retrospective observational comparative study.Patient's consent were not sought after as no patients' identifiers revealed to the public and no patient's radiological imaging with identifies were used.Patient's confidentiality was maintained in accordance with the declarations of Helsinki.The study was conducted in accordance with the requirements of the biomedical research ethics of the university and the provincial ethics board.However, there was no association between race and hemorrhage (p = 0.2).These results are not meaningful since the frequency between the races is uneven.

Table 1 summarizes the demographic and angioarchitectural
Of the 71 patients who presented with hemorrhage, 66 (92.9%) patients had a single hemorrhagic event and recurrent hemorrhage was found in 6 (8%) of the patients (not in the table).The frontal (20.4%) and parietal (22.3%) lobes were the most common locations.Half (49.7%) of the brain AVM cases were less than 3cm in maximum nidus diameter and only 12.7% were larger than 6 cm.Fifty-two percent of the brain AVMs had only superficial venous drainage, 22% had exclusive deep venous drainage and 26% being drained by both deep and superficial drainage.
The results in table 1 indicate that there was a strong association between brain AVM hemorrhage and AVM location (p = 0.001), nidus size (p = 0.002), venous drainage (p = 0.005) and the presence of associated intracranial aneurysms (p = 0.002).Intranidal aneurysms were the type associated with significant risk of brain AVM hemorrhage (p = 0.008).The Spetzler-Martin grading was not associated with brain AVM hemorrhage (p = 0.4).However, the analysis done for the results in Table 1 may be better interpreted using a modeling approach.Eleven (7.0%) patients had missing information regarding their management.Of the 146 patients with documented management information; medical, endovascular, surgical and stereotactic radiosurgery represented 67 (42.7%), 51 (32.5%), 13 (8.3%) and 15 (9.6%), respectively.Of the six patients who presented with recurrent hemorrhage, two patients had recurrent hemorrhage prior to interventional therapy, one after surgical excision and three following stereotactic radiosurgery.The median (IQR) follow up was 36 (12-72) months.
Table 2 presents the results for the simple and multiple logistic regression analysis with angioarchitectural features and demographic variables as the predictor variables and hemorrhagic presentation as the response variable.The univariate analysis using simple logistic regression found a significant association between brain AVM hemorrhage and AVM size <3 cm in maximal diameter (OR = 2.0; 95%CI: 1.1-3.8),exclusive deep venous drainage (OR = 3.9; 95%CI: 1.7-9.1),deep location (OR = 5.2; 95%CI: 2.4-10.9),and the presence of associated arterial aneurysms (OR=3.2;95%CI: 1.5-6.7).No significant association was found between brain AVM hemorrhagic presentation with age, gender, race/ethnicity and Spetzler-Ponce grading.
Multiple logistic regression analysis found deep location (OR = 3.3; 95%CI: 1.4-7.7)and the presence of arterial aneurysms (OR = 2.8; 95%CI: 1.4-6.6)as significant predictors of hemorrhagic presentation when compared to respective references.A borderline association was found between deep venous drainage and the increased risk of brain AVMs hemorrhage, in comparison to the superficial venous drainage, p = 0.06.Though the odds of brain AVMs < 3cm in maximal diameter of presenting with hemorrhage were high, compared to the lesions > 3 cm in size, there were no statistical significance (OR = 1.5; 95% CI: 0.7-3.4).

Discussion
The most serious clinical manifestation of brain arteriovenous malformation is intracranial hemorrhage [1,3,10].The ability to identify patients at risk of brain AVM hemorrhage, and the angioarchitectural features that predispose the brain AVM lesion to bleed will enable the treating physician to offer appropriate

ISSN: 2373-8995
therapy to patients whom interventional therapy is warranted.This study identified several angioarchitectural factors associated with brain AVM hemorrhage and these factors can serve as therapeutic target to prevent/reduce the risk of brain AVM hemorrhage.
Patients' age was found not to be a risk factor for brain AVM hemorrhage in our study.Our findings are consistent with the study by Stefani, et al. [21], who could not demonstrate the significant effect of age on brain AVM hemorrhages.However studies by Stapf, et al. [8] and LV, et al. [22] found the risk of hemorrhagic AVM presentation to differ significantly between age classes.Stefani, et al. [8] found that, highest hemorrhage frequency were detected among patients less than 10 years of age (57%) and those above 50 years of age (58%).Kim, et al. [4] reported a 30% increase in the risk of hemorrhage for every decade increase in age.In our study of 157 patients there were four patients younger than 10 years and three of them had bled, and there were 28 patients older than 50 years of whom 11(39.3%)presented with hemorrhage.
Our study found no differences in brain AVMs hemorrhage between male and female.Yamada, et al. [23] had reported borderline association between female and risk of brain AVM hemorrhage.Female patients in our studies harbored 1.4-fold risk of presenting with hemorrhage when compared to the male counterparts, though this result was not statistically significant.Our findings were consistent with those by Gross, et al. [14] who found no gender predisposition to hemorrhagic presentation.
In this study, we could not determine any association between race/ethnicity and the risk of brain AVMs hemorrhage.Though the odds of African-black patients of presenting with brain AVM hemorrhage were high compared to White patients, the difference was not statistically significant.The smaller sample size could be attributable to this insignificant difference.In addition, these results are not meaningful since the frequency distribution across the race groups is uneven.A recent large study by Kim, et al. [24] found a statistically significant increased risk for subsequent AVM hemorrhage among Hispanics compared with Whites (HR 3.1, 95% CI 1.3-7.4),and a non-significant trend for blacks and Asians (HR 2.1 and 2.4, respectively).The study by Kim et al shares similar findings with our study of nonsignificant increased risk of brain AVMs hemorrhage in Africanblacks compared to Whites.
Deep AVM location was found to be apredictor of brain AVMs hemorrhage, even after adjusting for size, venous drainage, arterial aneurysms, age and gender.Several studies have identified deep location to be a predictor of both initial and future hemorrhage [17,25,26].They suggest high frequency of perforator's supply [17], exclusive deep venous drainage and kinking of venous system at the level of the tentorium.Deep located brain AVMs does not present with seizures and only comes to clinical attention after hemorrhage [25].Marks, et al. [27] found no preferential location to presentation with hemorrhage.
Several studies have consistently reported the deep venous drainage to be associated with increased risk of hemorrhage in adult AVMs [25][26][27][28].Duong, et al. [13] suggested that deep venous drainage might have profound effect on brain AVMs hemodynamics by promoting an increased pressure gradient across the vasculature of the nidus.Kellner, et al. [2] postulated that deeply located brain AVMs typically have a smaller diameter and the deep draining veins, as well as high frequency of stenosis.All these factors make brain AVMs with exclusive venous drainage susceptible to rupture.Our study could not identify exclusive deep venous drainage as a significant predictor of brain AVM hemorrhage after adjusting for other variables in the model.However, though not significant, the odd ratio is large, and thus if our sample size was large, our result could have matched that found in the simple analysis and those reported by the previous studies [2,3,[25][26][27][28].
Our study had identified in the univariate analysis a significant association between brain AVM size < 3 cm in maximal diameter with the increased risk of hemorrhagic presentation.Small AVM are thought to be associated with a higher feeding mean artery pressure [29], which is known to be associated with increased risk of hemorrhage [13].Our univariate analysis results were in agreement with those by Ellis, et al. [25] and Spetzler, et al. [29] who found strong association between small AVMs ≤ 3cm in diameter and the increased risk of hemorrhage.Multiple logistic regression analysis however could not determine brain AVM size as a predictor of AVM hemorrhage when adjusted for age, location, venous drainage pattern and the presence of arterial aneurysms.This result is in agreement with some previous studies which could not determine AVM size as a significant predictor of brain AVM hemorrhage [21,26,27].
The presence of intracranial aneurysms in our study was associated with 3-fold increased risk of brain AVMs hemorrhage even after adjustment for other variables in the model.Intranidal aneurysms were the type of aneurysms with increased predisposition, while no significant association was found between flow-related aneurysms and the AVM lesions without aneurysms.Several studies have established the association between aneurysms and the risks of both initial and future brain AVMs hemorrhage [26][27][28]30].Da Costa, et al. [28] and Brown, et al. [30] had reported the annual risk of hemorrhagic presentation to be 7% per year at 5 years after diagnosis compared to 1.7% per year for group without aneurysms.This association was not found by other studies [13,21].

Conclusions
Brain AVM hemorrhage can result with a devastating consequences particularly in the developing world where resources for care and rehabilitation are scarce.The Timeous treatment of the patients at high risk of brain AVM hemorrhage will reduce the burden associated with the management and rehabilitation of the hemorrhagic stroke.By identifying factors that predispose brain AVMs to an increased risk of hemorrhage will help in timeous institution of therapy to patients whom interventional therapy is warranted.The results of this study have identified deep nidus location and the presence of aneurysms as the significant predictors of brain AVM hemorrhage.Although exclusive venous drainage and small nidus size < 3 cm were reported by other studies as the predictors of brain AVM hemorrhage, this study could not found that association.Small size used in this study could have been responsible for this lack of association.

Limitations
Our study is a retrospective study which relied on information already captured in patient's clinical records.The angioarchitectural features described are those of the lesion already ruptured which could be different to the lesion prior to rupture.The presence of hematoma and the mass effect it produces may influence the detection of some important angioarchitecture features such as aneurysms.The study was a hospital based and not a population based study, therefore it reported on brain AVM patients who already presented with hemorrhage.It does not account for people who may have similar but asymptomatic lesions in the community, who did not come in contact with health care system.The small sample size may have biased some of the results, therefore a future large multicenter study is recommended.

Table 1 :
Summary statistics of the demographic and angioarchitecture features.

2373-8995 features
of patients who present with or without brain AVMs hemorrhage.Our search identified 157 patients with brain AVMs.Seventy-one (45.2%) patients out of 157 presented with hemorrhage.The Median age was 31(IQR: 22-44) years.There was no significant difference in median ages between those who present

Table 2 :
Unadjusted and adjusted logistic regression analysis.