The influence of age and the initial clinical presentations of patients with an arteriovenous malformation on the risk of hemorrhage
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0028-3886.178047
Source of Support: None, Conflict of Interest: None
Background and Purpose: The major determinants of hemorrhage in an arteriovenous malformation (AVM) include its angioarchitecture, as well as the patient's age and clinical presentation. The aim of this study was to determine whether or not the risk of hemorrhage in an AVM differs based on the different clinical presentations and the initial age at diagnosis.
Keywords: Age; arteriovenous malformation; brain; hemorrhage
The angioarchitecture and the initial clinical presentations of brain arteriovenous malformations (BAVMs) may vary with the patient's age. ,, An increasing age may be an independent predictor of subsequent hemorrhage, feeding artery aneurysms, the diameter of a BAVM and venous ectasia. However, the influence of the patient's age and that of the clinical presentation of patients at the initial detection of the AVM, on the natural history of the AVMs has not been conclusively explored. We undertook this research to ascertain the relationship of the patient's age at the initial diagnosis and the initial clinical presentation with the risk of occurrence of subsequent hemorrhage within the AVM.
The medical records of all consecutively enrolled patients with an AVM who were admitted between the years 2004 and 2013 in the Department of Interventional Neuroradiology at the Beijing Tiantan Hospital, Capital Medical University, including their diagnosis, history, demographic features, clinical examination, and radiological information, were retrospectively reviewed. A total of 496 consecutive cases (280 male and 216 female patients) having a brain AVM were collected. The AVMs were initially detected utilizing a computed tomography (CT), magnetic resonance imaging (MRI), and angiographic examinations. Their age at the initial diagnosis ranged from 3-66 years (with the mean age being 26.8 years) and the initial presentations were: Hemorrhage in 264 (53.2%) cases, seizures in 88 (17.7%) cases, headache in 68 (13.7%) cases, and neurological deficits in 37 (7.5%) cases. The AVMS were detected incidentally in 39 (7.9%) cases. The Spetzler-Martin grades  of the AVMs included were: Grade I in 53 (10.7%) cases, grade II in 181 (36.5%) cases, grade III in 168 (33.9%) cases, grade IV in 83 (16.7%) cases, and grade V in 11 (2.2%) cases.
The first endpoint of the study was the occurrence of the subsequent hemorrhage after the initial diagnosis of an AVM had been established. Another endpoint for inclusion in this study was the initiation of any specific treatment for the cerebral AVM obliteration (including AVM resection and/or its embolization). The placement of an intraventricular drainage and a hematoma evacuation after the occurrence of hemorrhage in an AVM was, however, not considered as an end point for the termination of this study. The third endpoint was the last available date of follow-up prior to the initiation of any therapy for the obliteration of the AVM. This study did not require a formal ethical approval by the research ethics committee because of the retrospective nature of the study and the fact that the study relied on information obtained as a part of the routine clinical work. The duration from the initial diagnosis to enrollment of patients in our institute was 1-2 months in patients who initially presented with hemorrhage, and much longer in those patients who had other intial clinical presentations or those in whom the AVM was detected incidentally. The following variables were analyzed on our study: The initial presentation, the patient's age at the initial diagnosis of the AVM, the patient's gender, the Spetzler-Martin grade of the AVM, and the modified Rankin scale score (mRS) of the patient prior to and after the occurrence of hemorrhage. To investigate the association between the patient's age and the risk of occurrence of subsequent hemorrhage, the patient's age at the initial diagnosis was further categorized as <10 years, 10-19 years, 20-29 years, 30-39 years, 40-49 years and ≥50 years, based on our previous study.  The follow-up duration was calculated based on the date at which the initial diagnosis of AVM had been established to the end point of the study (the occurrence of a subsequent hemorrhage; the institution of a therapeutic intervention to excise or obliterate the AVM; or, the last available follow up before any intervention had been performed). The odd's ratio (OR) and 95% confidence intervals (CIs) were calculated using the univariate analysis to determine the risk factors responsible for a subsequent hemorrhage in a previously detected AVM. The significance of group differences was assessed using log-rank test of the Kaplan-Meier curves. The analyses stratified by each variable (initial presentation, age at initial diagnosis, gender and Spetzler-Martin grade) that assessed the influence of these factors on the likelihood of a subsequent hemorrhage in an AVM, were also conducted by the Kaplan-Meier curves. Statistical significance was determined at P < 0.05. Analyses were carried out using the statistical software SSPS for Windows (version 16.0; SPSS Inc., Chicago, IL).
Clinical presentations and age
There was a significant difference in the age of patients whose initial presentation was in the form of seizures when compared to those whose initial presentation was headache (mean, 25.5 years vs. 30.4 years, P = 0.013) and those in whom the AVM was detected incidentally (mean, 25.5 years vs. 26.8 years, P = 0.042). No significant difference, however, was noted in the patients who presented with seizures and those who had a neurological deficit at the initial presentation (P = 0.237). There was also no difference in the mean age (mean, 26.3 years vs. 27.3 years, P = 0.262), gender (P = 0.713), and Spetzler-Martin grades (P = 0.112) of patients who had an initial presentation of hemorrhage and those who did not.
Clinical presentation, age, and subsequent hemorrhage associated with the the arteriovenous malformation
During the mean duration of 2.5 years (range: 1 month-28 years), 84 patients had a subsequent hemorrhage associated with the AVM that resulted in a 6.8% annual hemorrhage risk. The percentage of patients with the hemorrhagic presentation rose from an initial figure of 53.2% to a subsequent figure of 58.7%. The morbidity (mRS ≥2) caused by the initial hemorrhagic presentation was 26.9%; the subsequent hemorrhage increased the morbidity by 9.5% (i.e., in 8 out of 84 patients, the mRS became >2). There was a significant difference in the mRS grade prior to and after the occurrence of the subsequent hemorrhage (mean mRS, 0.74 vs. 0.78, P = 0.017). However, there was no difference in the incidence of neurosurgical intervention in patients that was directed towards the initial and subsequent AVM hemorrhage (12.5% vs. 11.9%, P = 0.528).
In univariate analysis, the subsequent AVM hemorrhage was significantly associated with the initial hemorrhagic presentation (χ2 = 8.696, P = 0.003, OR = 1.548, 95% CI: [1.118-2.144]), a younger age (<30 years, χ2 = 9.555, P = 0.002, OR = 1.792, 95% CI: [1.186-2.707]), the female gender (χ2 = 4.133, P = 0.042, OR = 1.626, 95% CI: [1.015-2.606]), and the Spetzler-Martin grades III and IV (χ2 = 7.572, P = 0.006, OR = 1.964, 95% CI, [1.208-3.195]). In the patients presenting initially with hemorrhage, an initial diagnosis at a younger age was significantly associated with the occurrence of subsequent hemorrhage (mean, 22.3 years vs. 27.4 years, P = 0.007).
The mean follow-up period was 1.75 years (range 1month-28 years) in the hemorrhagic group and 3.3 years (range 1month-26 years) in the non-hemorrhagic group (P < 0.001). A subsequent AVM hemorrhage occurred in 57 patients during 462 patient - years among the hemorrhagic group, and in 27 patients during 774 patient - years among the non-hemorrhagic group. The annual bleeding rate after the initial detection of the AVM was 12.3% in the hemorrhagic group and 3.5% in the non- hemorrhagic group (P < 0.001, log-rank test) [Figure 1]a.
In the group that initially presented with hemorrhage, the annual hemorrhage rate in the patients with the age range 10-29 and 40-49 years was significantly higher than that for other age groups (P = 0.006, log-rank test) [Figure 1]b. In particular, the patients in the age range of 40-49 years had the highest annual hemorrhage risk of 22.4% when compared to the annual hemorrhage risk of 11.7% for other ages (P = 0.005, log-rank test). In the non hemorrhagic group, however, patients in the age groups <10 years and ≥50 years had a significantly higher risk of a first hemorrhage (P = 0.040, log-rank test) [Figure 1]c. Considering the initial presentation in the non-hemorrhagic group, the annual bleeding rate among those patients who initially presented with headache (annual bleeding rate of 4.9%), those in whom the AVM was detected incidentally (annual bleeding rate of 6.0%) and those who had initial neurological deficits (annual bleeding rate of 3.2%), was significantly higher than that seen in patients whose initial presentation was in the form of seizures (annual bleeding rate of 1.9%; P = 0.024, log-rank test) [Figure 1]d.
This is one of largest series reporting on the natural history of brain AVMs after their initial diagnosis and prior to any specific treatment being instituted for obliterating them . In our series, the mean time from the initial diagnosis of the AVMS to the institution of treatment was two times more in the patients in whom there was no initial hemorrhage as compared to those in whom hemorrhage was the initial manifestation. Those patients with AVMs in whom headache is the initial presentation or those in whom the AVM is found incidentally should also be under surveillance for their slightly higher subsequent hemorrhage rate than that seen in patients who present with seizures as their initial manifestation. This is contradictory to the previous opinion that the risk of hemorrhage in an asymptomatic AVM was low. Moreover, this study showed that there is a bimodal distribution of age at which an AVM is likely to rupture. In patients with an AVM whose initial presentation was hemorrhagic, the two age ranges for a higher rate of AVM rupture include 10-29-year and 40-49-years of age; whereas, in the patients with initial non-hemorrhagic manifestations, the peak ages at which the AVM is likely to rupture are seen at <10-years and after 50-years of age. These findings could represent a biological difference in the AVMs presenting with hemorrhage versus those that are presenting with non-hemorrhagic manifestations; and, between those AVMs presenting at a younger age at diagnosis versus those presenting at an older age at diagnosis. An older age group may harbor additional risk factors such as concurrent arterial aneurysms, ,, the presence of which could have elevated the risk of hemorrhage.  The mechanism involved in the hemorrhage of AVMs may also differ in different ages. The AVM nidus and pseudoaneurysm rupture may occur more often in younger patients; and, a venous ectasia and an aneurysm on a feeding artery may rupture more frequently in older patients. These high-risk features could be targeted and treated to decrease the risk of hemorrhage.  The different mechanisms of AVM hemorrhage also resulted in hemorrhage at different sites. Thus, an associated aneurysmal rupture resulted in a subarachnoid hemorrhage;  a deep venous bleed predominantly caused an intraventricular hemorrhage; and, a hemorrhage within the AVM nidus caused a parenchymal hemorrhage. The clinical impact of these different mechanisms responsible for hemorrhage in an AVM is much more than that seen due to the usual factors responsible for intracranial hemorrhages in otherwise healthy brain tissue, such as hypertensive cerebral hemorrhage or cerebral amyloid angiopathy. 
Our findings suggest that AVMs in children do not appear to be at increased risk for a subsequent hemorrhage compared with adults, which is consistent with the findings reported by Fullerton et al., and Yamada et al. , Hetts et al., found that AVMs in children were more likely to rupture than adults, but venous ectasias and feeding artery aneurysms, which are particularly high-risk features and take time to develop, were under-represented in children.  The results of the latter study may have been confounded by the fact that hemorrhagic presentation was higher in children simply because unruptured AVMs in them were less often detected.
The presentation of seizures in AVMs may help in focusing attention on their underlying existence and often influence the decision-making regarding their treatment. Observational case series describe encouraging differences in seizure frequency before and after AVM treatment,  but the shortage of studies demonstrating dramatic effects in comparison to the control groups justifies the need for more controlled studies, ideally with randomized treatment allocation.  Interventional treatment of AVMs may have an impact on the clinical outcome by attenuating the risk of re-bleeding but may also be harmful and result in the poor functional outcome and even death. ,,,,, A prospective population-based study suggested that there was no difference in the 5-year risk of seizures with or without treatment of the underlying AVM, irrespective of whether the AVM had presented with hemorrhage or epileptic seizures.  Our study provided an annual hemorrhage rate of 6.0% for incidental AVMs and 4.9% for AVMs associated with headache, which is similar to the results of 6.44% and 6.48, respectively reported by Yamada et al.  Incidentally detected AVMs may present with various non-specific complaints including mild headache, trigeminal neuralgia, or dizziness, that may focus attention on the underlying pathology. The results of our study suggested that incidental AVMs and initial headache episodes associated with AVMs should also be observed carefully.
Our study was a hospital-based cohort and showed a moderate to severe disability (mRS ≥2) in 26.9% of the patients. The difference between the mean modified Rankin score at the initial diagnosis and at follow-up after the occurrence of the hemorrhagic episode was significant. In another hospital-based cohort, the investigators found no disability (mRS 0 or 1) in 49% patients, and a moderate to severe disability (mRS ≥2) in 51% patients, but no patient died after the initial AVM hemorrhage. , In contrast, in a population-based design, the 30-day case-fatality rate was found to be 11%, and 38% of the patients in whom there was an associated hemorrhage within the AVM were either dead or dependent after 1-year.  This discrepancy may be caused by a referral bias. Our center is a tertiary referral center so that referral bias may have affected our results. Furthermore, results might have been confounded by the recovery of neurological deficits before the referral. The hemorrhage associated with an AVM also included subarachnoid and/or intraventricular hemorrhage, which have a different prognosis when compared to a parenchymal hemorrhage.
The results of our univariate analysis suggested that female gender and Spetzler-Martin III and IV grades were independent risk factors for the rebleeding of cerebral AVMs. The female gender and exclusively deep venous drainage were additional predictors of hemorrhage reported in a meta-analysis.  Pregnancy and the postpartum period may have an important role in influencing the rebleeding rates in an AVM that may increase the incidence of hemorrhage in the female patients. A recent study, however, found that the bleeding rates in an AVM did not increase in patients with a cerebral AVM during pregnancy and puerperium. , Stapf et al.,  reported that no association existed between the female gender and the subsequent rate of hemorrhage of an AVM. Spetzler-Martin III and IV grades were prone to developing a hemorrhage when they were of a small size, had a deep-seated location, and had a deep venous drainage. ,, However, untreated Spetzler-Martin grade IV and V presenting with hemorrhage also show a significant risk of subsequent rupture. Their subsequent rupture carries a higher risk of developing case fatality and permanent morbidity than are seen in AVMs in general.  The AVM size was not associated with an increased risk of developing an intracerebral hemorrhage in multivariable models. 
The goal of AVM treatment should be the prevention of hemorrhage. Microsurgery (i.e., craniotomy followed by resection) has been reported to have a low risk of complications for Spetzler-Martin grades I and II brain AVMs (e.g., small malformations in noneloquent areas) and results in an immediate cure. However, microsurgery is invasive. Stereotactic radiosurgery is effective for malformations that are smaller than 3.5 cm in size, but complete obliteration requires approximately 1-3 years after treatment and cure is not always obtained. Delayed complications such as hemorrhage in the latency period, and radiation induced edema or necrosis can occur. Embolization is used to obliterate small malformations or to make larger malformations amenable for radiosurgery or surgery, or to eliminate a possible cause of hemorrhage (e.g., associated aneurysms). For embolization, microcatheters are used to deliver embolic material to the feeding arteries or the nidus of the AVM [Figure 2]. Brain AVMs categorized as either Spetzler-Martin grade IV or V generally require multimodality treatment. Some of the prominent studies related to the treatment and outcome of patients harboring an AVM are listed in [Table 1]. ,,,,,,,,,,,,,,,,,,,,,,,,,,,, The morbidity and mortality rate was 10-20% after microsurgery, and 0.7-19% after embolization. However, the long-term follow-up of untreated cases showed a much higher 30% morbidity and mortality rate.
In our department, the majority of patient underwent endovascular management of their cerebral vascular diseases, including brain AVMs. The potential bias included the selection of patients in a relatively better neurological grade that had the potential of lowering the observed overall annual hemorrhage rate. In the present series, the annual hemorrhage risk was 12.3%/year, which was higher than the previously presented data in other studies.  It is likely that this higher incidence in the present study is due to the shorter duration of follow-up. Despite the fact that this study is composed of patients from a single center, it recruits a very large number and is very homogeneous in terms of the patient characteristics.
The annual hemorrhage rate of AVMs after their initial diagnosis is significant for patients in whom an initial hemorrhagic presentation was present and in patients between the age ranges of 10-29 years and 40-49 years. The risk of subsequent hemorrhage in patients in whom the initial presentation was in the form of seizures was lesser than that seen in other presentations. The patients in whom the AVM presented with headache, and those in whom the AVM had been detected incidentally should also be considered for surveillance.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
[Figure 1], [Figure 2]