Preliminary Results of the ARUBA Study

Arteriovenous malformations (AVMs) are congenital lesions that, when left untreated, portend a lifelong cumulative risk of stroke and death. The fact that eradicating an AVM to eliminate its natural risk comes at a price to the patient is not news and does not warrant investigation. What is worthy of scrutiny is the need to value and scale this induced morbidity in the light of the gain achieved, ie: a life with long lasting cure. ARUBA tells us what we already know: there is an initial price attached to the intervention. It does not tell us whether the price is too high for any specific AVM. Yet, in spite of the grave methodological shortcomings of the trial that highly bias its results from the outset against intervention, intervention still emerged a superior option for the Spetzler-Martin Grade 1 AVM. The message is clear. There never was clinical equipoise to justify randomizing grade 1, and almost certainly grade 2 patients. Equally, there never was equipoise for the enrolled grade 4 patients, who should be, by and large, left untreated. A registry of at least all unselected grade 3 patients is what is needed to evoke meaningful data that can preserve external validity.

On May 10, 2013, the National Institute of Neurological Disorders and Stroke (NINDS) announced that A Randomized Trial of Unruptured Brain AVMs study (ARUBA) had prematurely stopped enrollment—a result of the pre-planned interim analysis performed by the trial’s independent Data and Safety Monitoring Board (DSMB), which demonstrated an event rate three times higher in the intervention group than in the medical management group.¹

ARUBA was a randomized, multi-center trial comparing “best possible AVM eradication” with non-invasive, medical management to the primary endpoint, a composite of symptomatic stroke or death. Methods to eradicate brain arteriovenous malformations (BAVM) included radiosurgery, microsurgical resection, and endovascular embolization, alone or in combination. The secondary outcome measure was disability as measured by the Rankin Score at 5 years post-randomization. The initial study design had called for enrollment of 800 patients, but this number was later reduced to 400 patients after an interim sensitivity analysis. The final number enrolled became significantly smaller still, with only 223 patients at the time enrollment was halted. Patients with previous BAVM hemorrhage or treatment, and those with BAVMs considered untreatable for complete removal or eradication, were excluded from the trial. As a result of the decision by the trial’s DSMB, the study investigators stopped enrollment but will continue to follow patients already enrolled.

On May 31, 2013, the preliminary results from ARUBA were presented at the European Stroke Conference in London, England. In summary form only, the study was reported to have enrolled 223 patients from 65 certified sites. A great majority of the patients were recruited from centers not in the United States, with slightly more than 40 patients enrolled in the United States. One hundred twenty six patients were managed medically while 97 were treated with interventional modalities. Most patients presented with symptoms of seizure (43%) or headache (52%) while focal deficits were rare (14%). Ninety-four patients (42%) were asymptomatic at the time of diagnosis. BAVMs in the study cohorts were relatively well matched for size, location, and venous drainage pattern as well as for Spetzler-Martin (S-M) Grade. The majority was comprised of S-M grade 1, 2, or 3 (approximately equal) with a few grade 4 patients (10%) and no grade 5 patients enrolled. The average follow-up period was 33 months.

In an analysis based on the treatment received, the authors report that ten adverse events (7.9%), defined as death or stroke, occurred in the medical management arm while 34 adverse events (35.1%) occurred in the interventional therapy arm (RR = 0.35, 95% CI = 0.19-0.65). There was no significant difference in the number of deaths (two in the medical management arm vs three in the interventional arm). When analyzed according to S-M Grade, patients with grade 1 BAVMs fared better in the interventional group while adverse events were much higher in patients with grades 2 to 4 lesions. Of interest, 45% of patients assigned to interventional therapy in the United States were still awaiting treatment at the time the study was halted. In patients followed for 24 months (n = 123), the modified Rankin Score was >2 (secondary endpoint) in 36.1% of patients in the interventional arm and 9.7% in the medical management arm. No specific data was provided in the presentation about methods used to treat patients in the interventional arm (radiotherapy vs endovascular vs microsurgery vs combination therapy), and the trial was not powered to evaluate treatment effect by modality.

While intending to shed light on the best management of unruptured BAVMs, the ARUBA trial has not been free of criticism. Multiple difficulties with the study have been previously described.^2,3 Among these are the lack of physician accreditation to participate in ARUBA beyond a statement that more than ten BAVMs are treated annually at participating centers. This diverges from more rigorous physician accreditation required by other recent trials investigating comparatively homogeneous cerebrovascular disorders, such as those performed for the management of carotid stenosis and intracranial stenosis.^4-6 Clear differences in end-point morbidity exist between radiosurgical intervention compared with endovascular or microsurgical interventions, yet the trial was not powered to examine these treatment-specific variations. While some cerebrovascular centers have extensive experience in the treatment of BAVMs, inclusion of centers without extensive experience is problematic. Finally, initial surgical morbidity in a study with short-term follow-upsuch as ARUBA can have a profound effect on outcomes, particularly when compared to medical management in a lesion with a relatively low but life-long annual risk of hemorrhage or disability.

A major limitation of studies such as ARUBA is selection bias. External validity of ARUBA (applicability to BAVM patients in general) depends on the sample population of the study patients. Given enrollment of only 223 patients, compared with the greater number of BAVMs actually evaluated, only certain patients were selected for trial participation. Treatment of BAVMs has evolved along several well-established patterns of therapy. It is quite possible that those patients thought to be at high risk for BAVM rupture without treatment by study investigators were excluded from enrollment and treated outside of the confines of the trial. The International Study of Unruptured Intracranial Aneurysms (ISUIA) study demonstrated similar limitations as high-risk aneurysms were, by their very nature, pre-selected out of the study, leaving a group of patients enrolled in ISUIA with a lower risk for aneurysmal rupture.^7,8 Lack of equipoise and the discomfort with natural history risk among clinicians who treat large numbers of BAVMs in tertiary care centers in the United States has likely resulted in the low number of participating US centers as well as the low study accrual rate.

Moving forward, the ARUBA investigators are requesting additional funding to continue monitoring the 223 enrolled patients, potentially for up to 10 years. On first inspection, this request seems like an appropriate response to criticisms that the existing trial follow-up is too short. However, ARUBA was already enormously expensive and it is important to carefully assess the usefulness of such a decision.

Longer-term follow-up could potentially show equality in outcomes or even the superiority of intervention as untreated patients in the medical arm of ARUBA experience clinical events. Similarly, treated patients in the interventional arm commonly demonstrate progressive neurological improvement over time, at least for the secondary outcome measure (modified Rankin score). However, such a reversal will not alter the external validity of the trial. A crossover of outcomes such that patients managed non-invasively suffer a higher incidence of stroke and death than those treated should not lead to changes in management algorithms: even then it would be inappropriate to conclude that all persons with unruptured BAVM should necessarily be treated. We feel that the next most appropriate step in the investigation of management of unruptured BAVMs would be to create and promote a long-term, comprehensive, and adjudicated registry inclusive of all BAVM patients.

In conclusion, the preliminary data presented from the ARUBA trial are not surprising. Any randomized trial attempting to determine management paradigms for lesions as heterogeneous and rare as BAVMs is limited in the interpretation of its results and in its external validity. The trial results reinforce our knowledge of the high complication rates for interventional treatment of high-grade (S-M 4-5) BAVMs as well as our knowledge of the safety of intervention for grade 1 lesions. Selection bias in ARUBA may make it impossible to extrapolate any of its conclusions to the management of all unruptured BAVMs. The small number of study patients will also make post hoc analysis of specific subgroups problematic. In light of these limitations, it may be more productive to study such rare and heterogeneous lesions using multicenter, adjudicated, consecutive patient registries to capture all patients without bias.