Pathogenesis and Radiobiology of Brain Arteriovenous Malformations: Implications for Risk Stratification in Natural History and Post-treatment Course

Achal S. Achrol, B.S.; Raphael Guzman, M.D.; Monika Varga, B.S.; John R. Adler, M.D.; Gary K. Steinberg, M.D.; Ph.D.; Steven D. Chang, M.D.


Neurosurg Focus. 2009;26(5):E9 

In This Article

Clinical Characteristics and Natural History: A Need for Biofeedback

Brain AVMs can present with ICH or they can be unruptured, that is, with symptoms resulting from mass effect, inflammation, altered hemodynamics from direct shunting, or recruitment of dilated perinidal capillary networks (for example, progressive neurological deficits, seizures, recurrent headaches, and congestive heart failure in neonates).[9,21,27,31,32,51,63,72,80,98]

Following the advent of noninvasive imaging techniques, including the widespread availability of CT and MR imaging, there has been a significant increase in the incidental detection of unruptured BAVMs,[10] which until now represented only a minority of BAVM diagnoses in previous studies.[21,72] In most recent studies, however, unruptured BAVMs now account for the majority of cases.[51,89]

Some investigators have expressed concerns over the optimal management strategies for the growing number of patients harboring unruptured lesions,[90] and the role of prophylactic intervention in unruptured BAVMs is the subject of ARUBA.[69] These investigators pointed to the classic study of Ondra et al.[72] as an example of insufficient evidence to support treatment of unruptured BAVMs.[90] This study reported annual hemorrhage rates of 2-4% in patients with untreated BAVMs (major morbidity 1.7%, mortality 1%, and combined morbidity and mortality 2.7% per year); however, the vast majority (71%) of these BAVMs had ruptured at initial presentation. As such, some investigators argue that these data actually blend estimates of recurrent hemorrhage risk from a majority of ruptured BAVMs with estimates of new ICH risk from a minority of unruptured BAVMs. These investigators hypothesized that, as a result, classic data likely overestimate new ICH risk among unruptured BAVMs and underestimate recurrent stroke risk among ruptured BAVMs. Similar issues have been raised regarding the classic analysis by Crawford et al.[21] of 343 patients with BAVMs treated without surgery; they were found to have experienced a 42% risk of hemorrhage, a 29% risk of death, and a 27% risk of neurological morbidity by 20 years after diagnosis, as the majority of this cohort (64%) had ruptured BAVMs at initial presentation.

Nevertheless, data are lacking to support the hypothesis that differences exist between classic AVM cohorts and modern-day cohorts in terms of overall annual ICH risk estimates or that increasing numbers of unruptured BAVMs are implicated in changing trends across studies over time. Two large, modern North American cohorts[51,89] encompassing 2086 patients with BAVMs and reflecting a preponderance of unruptured lesions (53-55%) have independently estimated the annual ICH risk to be 2.1-2.8% overall (1.3-1.4% for unruptured lesions, 3.7-5.9% for ruptured lesions at initial presentation).[51,89] This yields overall 20-year cumulative ICH estimates of 43-56%—that is, identical to or greater than what was observed in the study by Crawford et al.[21] despite a 24% decrease in the proportion of hemorrhagic presentations over time. This would seem to argue against the hypothesis that previous studies overestimated the annual risk due to a larger proportion of ruptured lesions.[21,72] With a mean age of 36 years among patients with BAVM in these 2 modern studies,[51,89] and assuming a life expectancy of 44 years by 2004 Social Security Actuarial Publications as previously described,[20] the cumulative lifetime ICH risk from unruptured BAVMs in the patients in these studies is estimated to be 59%, with a 5-year stroke risk of 6.8%. In one of these cohorts, 33% of patients who suffered new ICH demonstrated poor neurological outcome on 30-day modified Rankin Scale assessment (score > 2), and 18% of these individuals went on to suffer recurrent hemorrhage, leaving 47% with poor neurological outcomes.[18] Therefore, these data suggest that unruptured BAVMs carry a significant risk of future ICH and associated morbidity, albeit less dramatic than in ruptured BAVMs.

Given that patients with BAVMs are at lifelong cumulative risk of future ICH and typically receive the diagnosis at a young age, it is the subject of much controversy whether conclusions drawn from a 5-year study period like that of the ARUBA trial will generate accurate data of all risks and benefits comparing surgical prophylaxis against a lifetime of natural history risk from an untreated, unruptured BAVM.[20,64] In the estimates above, it is clear that a 5-year cumulative stroke risk of 6.8% would appear at best to show no surgical benefit when weighed in the short term against acute surgical morbidity associated with invasive therapy. However, this would be ignoring the feared lifetime cumulative stroke risk of 59% that forms the true motivation for surgical intervention.

More recently, the AVM database of Helsinki University in Finland has provided compelling new data on the natural history of BAVM.[43,56] Reviewing > 60 years of experience, these authors identified 631 consecutive patients with BAVMs.[43] The majority of these patients, however, were excluded from the natural history analysis, which was limited to patients with at least 1 month of hemorrhage- and treatment-free follow-up.[43] Nevertheless, the resulting group of 238 patients followed up for a mean duration of 13.5 years provides important data on rupture rate of untreated AVMs, even though the majority of these lesions were previously ruptured (58.4%).[43] These authors confirmed that the annual hemorrhage risk from an AVM was 2.4%, noting that the rate was highest during the first 5 years after diagnosis (4.7%) compared with > 5 years after diagnosis (1.6%).[43] This effect was largely driven by the majority of previously ruptured lesions in this cohort, which had a 6.2% annual hemorrhage risk during the first 5 years compared with 2.3% for unruptured lesions.[43] Importantly, these authors have also demonstrated that excess mortality from BAVM compared with the normal population was highest if the BAVM was treated conservatively, intermediate if the BAVM was partially occluded, and lowest if the BAVM was completely occluded.[56] Once the BAVM was occluded, there was no ongoing excess death after the 1st year.[56] These authors noted that these benefits were seen 5-7 years after treatment[56] and, therefore, shorter-term studies such as the ARUBA study may not be adequate to show such a benefit.

For these reasons, it is unclear whether a consensus can emerge from current clinical trials and epidemiological studies alone. This is especially true in light of the complexity of BAVM covariates that can each contribute uniquely to ICH risk in a specific patient, including age, ethnicity, traditional cardiovascular risk factors (smoking, hypertension, hypercholesterolemia, and diabetes), AVM size, location, venous drainage, and presence of associated aneurysms. The complexity of the disease makes it unlikely that a generalized rule, such as to offer surgery or not, can be applied on the basis of a single clinical variable alone, for example, unruptured BAVM status.

On the other hand, research efforts focusing on biofeedback have the potential to uncover disease markers that overcome the limitations of population-based research in informing treatment decisions at a patient-specific level. Proteomic analyses of sera from patients with BAVM are a promising research tool to identify biological patient-specific markers of vascular remodeling and inflammation that may signal the presence of vascular instability in a patient before the onset of a hemorrhagic event. Two-dimensional polyacrylamide gel electrophoresis sorts proteins by isoelectric focusing and sodium dodecyl sulfate-polyacrylamide gel electrophoresis separates proteins by molecular weight.[4,5,25,60] These techniques can be coupled with mass spectrometers, including matrix-assisted laser desorption ionization time of flight mass spectrometers (MALDI-TOF-MS) and electrospray ionization tandem mass spectrometers (ESI-tandem-MS) to detect differentially expressed proteins in sera.[4,5,25,60] Protein biomarkers of interest identified by these techniques can then be validated using Western blot and immunohistochemical analysis.

In addition to proteomic approaches for serum biomarker detection, gene expression profiling of peripheral blood mononuclear cells represents another promising approach to detect patient-specific biological signals of vascular instability and inflammation by profiling immune signatures and circulating EPC expression patterns.[84,94,95]

These proteomic and gene expression profiling techniques may provide patient-specific data on circulating immune and endothelial progenitor cell activity that can provide clinicians early insight into the inflammation and abnormal vascular remodeling characteristic of unstable BAVMs. Such research efforts, focusing on patient-specific biofeedback as a means to stratify patients, may hold the greatest potential to address current controversies over the treatment of patients with BAVMs by offering new methods of risk stratification and treatment selection that overcome conclusions generalized from population-based studies.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.