8.1 Goals

1. 
   

   Analyze the literature regarding the natural history of arteriovenous malformations (AVMs).

   
   
2. 
   

   Assess studies that investigate treatment of unruptured AVMs versus observation.

   
   
3. 
   

   Briefly review AVM treatment options, risks, and success rates.

8.2 Case Example

8.2.2 Treatment Plan

The patient agreed to treatment of the symptomatic AVM. The recommendation was made for neoadjuvant embolization of the deep feeder to ease the surgical removal of the AVM. Postoperative angiogram seen in Fig. 8.2a, b demonstrates no residual AVM.

8.3 Case Summary

1. 
   

   What would you consider the rupture risk of the unruptured AVM in this patient?

   
   

   Our knowledge, with regard to the natural risk of AVM rupture, is limited to prospective and retrospective cohorts. Later in this chapter, we will review the largest of these studies and the strengths and limitations of each one. The most recent meta-analyses report the annual rupture risk for unruptured brain AVMs to be 1.3 to 2.2%, but the overall literature supports an annual risk of 1 to 4%.!, ²

   
   
2. 
   

   What patient factors would you consider when deciding on your recommendations for observation or treatment of these unruptured cerebral AVMs?

   
   
   
   
   1. 
      

      Age

      
      

      Since it is believed that the annual risk of hemorrhage for AVMs remains constant, younger patients have a higher lifetime risk of hemorrhage; therefore, there is more precedent for treatment. Conversely, younger patients may have to live longer with deficits incurred due to complications of an intervention. It is imperative to weigh the risks of treatment versus conservative management as we will discuss later.

      
      

      In elderly patients with low-risk lesions, it may be better to intervene as opposed to observation since increasing age has been identified as a possible risk factor for rupture in some meta-analysis. Other studies failed to reach the same conclusion, and this may be due to lead time, selection, and follow-up bias. ^{1 , 3 , 4 , 5 , 6}

      
      
   2. 
      

      Patient symptoms

      
      

      Other than hemorrhage, brain AVMs may present with a variety of complaints and neurological alterations. Often unruptured AVMs can cause focal or secondary seizures, with a reported 5-year risk of seizure estimated around 8%. ^{7 , 8} Younger age, temporal location, cortical involvement, and nidus diameter > 3 cm increase the risk of seizure. ^{3 , 8} It is not certain that intervention reduces seizure risk; however, a meta-analysis reported that microsurgery had the highest seizure reduction rate (78.3%), followed by stereotactic radiosurgery (SRS) (62.8%), and embolization (49.3%). ⁹

      
      

      Unruptured AVMs may present with headaches and other variable symptoms, such as focal weakness or visual alterations. There is no evidence showing that AVM intervention reduces headaches. ¹⁰ However, in our case example, the patient had almost complete resolution of symptoms.

   
   
3. 
   

   What AVM factors would you consider when deciding on your recommendations for observation or treatment?

   
   
   
   
   1. 
      

      Prior hemorrhage

      
      

      While our patient never suffered a hemorrhage, the largest risk factor for future hemorrhage of an AVM is a prior hemorrhage. ³

      
      
   2. 
      

      AVM location

      
      

      It has been shown that deep and infratentorial AVM locations are independent risk factors for hemorrhage. ^{1 , 2 , 3 , 4 , 5 , 6}

      
      
   3. 
      

      Various angioarchitecture characteristics

      
      

      There are several AVM angioarchitectural characteristics that increase its risk of hemorrhage. In short, venous outflow limited to deep venous structures, associated arterial aneurysms (pre- or intranidal), single draining veins, venous varices, and venous stenosis are all considered independent risk factors. ^{2 , 3 , 6 , 11 , 12 , 13 , 14} Our patient had a large perinidal aneurysm, increasing her risk for future hemorrhage.

   
   
4. 
   

   How would you follow this AVM with or without treatment? There are several methods for following treated AVMs with imaging studies: digital subtraction angiography, computed tomography angiography, or magnetic resonance angiography. Depending on the method of treatment, intensive imaging follow-up may be necessary. It has been reported that magnetic resonance angiography follow-up offers similar accuracy in lesions greater than 1 cm. ¹⁵ All patients should have an angiogram at some point to ensure complete obliteration. In this instance, we elected to follow up with digital subtraction angiography.

8.4 Level of Evidence

Patient’s age, symptoms, and AVM grade: Given the Spetzler-Martin grade IV AVM, surgical treatment poses significant risks; however, since the patient was young, had neurological deficit, and a perinidal aneurysm, we elected to proceed with intervention (Class lib, Level of Evidence B).

Risk of future hemorrhage: The patient was young and had a large perinidal aneurysm (Class Ha, Level of Evidence A).

Treatment: Due to the large size of the lesion, we decided against SRS treatment (Class I, Level of Evidence B). We decided to perform preoperative embolization to decrease the morbidity and mortality associated with the surgery resection (Class Ha, Level of Evidence B).

Follow-up of aneurysm: We followed this lesion with digital subtraction angiography (Class I, Level of Evidence B).

8.5 Hemorrhage Risk in Unruptured AVMs

8.6 Management: Conservative versus Intervention

In 2010, Ross and Al-Shahi Salman conducted a systematic review to identify randomized clinical studies comparing outcomes following different unruptured AVM management. ⁷ No completed study met their inclusion criteria; however, they identified the ARUBA study ¹⁷ as the only ongoing study investigating this important question.

MohrJP, Parides MK, StapfC, et al. Medical management with or without interventional therapy for unruptured brain arteriovenous malformations (ARUBA): a multicentre, non-blinded, randomised trial. The Lancet 2014;383:614-621.

From 2007 to 2013, Mohr et al screened 1,740 patients for trial eligibility, ultimately randomizing 226 patients to either intervention (n = 116) or medical management (n = 110). Patients were assessed for eligibility if they were older than 18 years, had neurovascular image confirmed AVM, and no prior hemorrhage or treatment attempt. Interventions included radiosurgery, microsurgery, endovascular intervention, or any combination of these procedures. Primary outcome was time until any-cause death or symptomatic stroke, defined as any new neurological focal deficit, headache, or seizure with associated image findings (blood or low-density ischemic lesions). Secondary outcome was functional dependency at 5 years, defined as modified Rankin scale of 2 or higher. ¹⁹

Originally designed to enroll 800 patients, sample size was reduced to 400 patients following slow patient accrual in the first 18 months. The ARUBA trial was prematurely halted when conservative management reached a predetermined efficacy, for the prevention of stroke or death, in an interim analysis. A total of 223 patients were randomized, with a mean follow-up of 33.3 months (IQR 16.3-49.8 months). The final statistical analysis was conducted with 53% of the anticipated data (of the 400-patient sample size). Under both intention-to-treat and as-treated analyses, investigators found that conservative management had a significantly lower risk for primary outcome event than intervention. In addition, the risk of death or neurological disability was significantly lower in the conservatively managed cohort. Further analysis demonstrated that primary event occurrence remained similar regardless of Spetzler-Martin (S-M) grade in the medically managed arm; however, it occurred more frequently in patients with an S-M grade II or III in the intervention arm.

The ARUBA study reached the controversial conclusion that the risk of death or stroke in patients with unruptured AVMs was higher for patients undergoing intervention versus conservatively managed for 33 months of follow-up. In 2017, Magro et al conducted a systematic review of all the published critiques and comments involving ARUBA and identified 31 published critiques. ²⁰ While the ARUBA trial addressed a paramount question, the trial design and possible biases limited the external validity of the study. Regarding study design, the academic community has critiqued the study’s end points, selection criteria, and absent treatment arm standardization.

Many stated that the end points were “soft” and favored conservative management. ²⁰ The primary end point of the trial labeled a stroke as any postoperative new or worsening neurological focal deficit, headache, or seizure with associated imaging findings (blood or low-density ischemic lesions). Practitioners argue that a headache or an initial worsening of symptoms with contrast leakage often transiently occurs following an endovascular intervention. This argument was supported by an unusually high primary outcome incidence of 30.7% in the intervention arm, most of which occurred during the first 6 months following intervention. Furthermore, the AVM obliteration rate was not reported—a relevant statistic considering that partial AVM embolization does not necessarily reduce hemorrhage risk. ³ Authors did not specify if endovascular embolization was performed with cyanoacrylate-based liquid embolic agents or Onyx, each of which have different success rates. ³

The ARUBA trial was also subjected to selection bias. During its enrollment period, investigators screened 1,740 patients, of which only 226 were ultimately enrolled. The remaining patients were split as followed: 1,014 were not eligible due to prior hemorrhage, 323 refused participation, and 177 patients had the clinician select treatment outside randomization. Most high-volume centers (treating more than 50 patients per year) enrolled one to two patients per year. It is imperative to appreciate these numbers when interpreting the results. For 177 patients to have been selected for treatment outside of randomization, it implies that for each of those patients the clinician felt uncomfortable to randomize the patient to conservative treatment. This calls into question ARUBA’s equipoise and leads to the next commonly discussed pitfall of ARUBA, lack of standardization of the intervention arm.

In the intervention arm, 66 patients received single treatment therapy; 30 had embolization, 31 radiosurgeries, and 5 microsurgeries. The remainder of the intervention arm received embolization prior to radiosurgery (n = 15), surgery (n = 12), and both surgery and radiosurgery (n = 1). Critics often point out that surgical therapy is underrepresented. ^{3 , 20} The reported rates of surgical risk with Spetzler-Martin grades I—II are around 0.7% and incur high obliteration success. ²¹ In most circumstances, experts feel that for grade I—II lesions in well-selected patients intervention is warranted and there is no equipoise regarding treatment. While we are uncertain of what treatment options were taken with the aforementioned 177 patients who were excluded from the study, due to the low surgical representation in ARUBA it is possible that these patients fell into this low surgical risk category, implying that the treating physicians felt ethically obliged to exclude them from randomization. Bervini et al’s analysis of their prospectively maintained cohort of 377 unruptured AVMs supports this presumption. ²² They found that the 5-year rupture risk was 11.5% for an unruptured AVM, whereas only 1.6% of the Spetzler-Ponce class A AVMs treated by surgery had a risk for a permanent neurological deficit leading to an mRS score > 1. Their results showed that surgical treatment is favorable in these low-graded AVMs.

One might argue that many of the complications following AVM intervention impose a “front-loaded” risk, or occur shortly following the surgery, whereas AVM hemorrhage risk is believed to be a compounding risk, or similar year-after-year risk or hemorrhage. This leads to another main pitfall of the ARUBA study, short follow-up time. Literature regarding unruptured AVM management with a mean follow-up of 33 months leaves room for misinterpretation of the main results. Even with the original planned follow-up for 15 to 20 years, the study may be underpowered as a pretrial analysis with 400 patients demonstrated that 10 to 20 years would be necessary before the intervention curve would cross the nonintervention curve. The ARUBA investigators plan on following the patients for five additional years to address this concern; however, a longer follow-up may be necessary to answer the question.

Some additional critiques of the ARUBA trial include the slow study enrollment and the premature interruption of enrollment. ²⁰ However, these two critiques are somewhat unavoidable. Trial investigators designed a prior safety threshold to halt the trial and patient accrual was slower than anticipated; thus, the study was repowered accordingly.

Regardless of the outcome and critiques, ARUBA investigators conducted an ethically difficult trial that challenged the standard of care, ultimately opening discussion to a crucial question that requires further elucidation. Often practitioners consider RCTs to be the highest level of evidence. However, in circumstances where ethics and equipoise make conducting RCTs technically challenging, observational studies may be conducted. While ARUBA has been the only RCT to date investigating conservative versus interventional management for AVMs, important observational studies have been conducted.

Al-Shahi Salman R, White PM, Counsell CE, et al. Outcome after conservative management or intervention for unruptured brain arteriovenous malformations. JAMA 2014;311:1661-1669.

One of the more influential observational studies, regarding AVM treatment, was published a few months after ARUBA by the Scottish Audit of Intravascular Malformations (SIVMS). Al-Shahi et al conducted a prospective observational study of 204 Scotland residents who were diagnosed with unruptured AVM between the years 1999-2003 and 2006-2010. ²³ Patients were included if they were 16 years or older with a radiographically confirmed unruptured AVM and no distant cerebral aneurysms. Interventions were any combination of embolization (glue or coil), radiosurgery, and microsurgery. Decision to pursue intervention or conservative treatment was left up to the patients and their treating physicians. Patients were followed from the time of diagnosis (conservative treatment) or from the time of treatment for the patients who underwent intervention. The primary outcome was first occurrence of handicap, defined as Oxford Handicap Scale 2-5, that was sustained for at least 2 years or any-cause death. Secondary outcome was death, due to intervention or AVM, or nonfatal symptomatic stroke.

During the study period, 101 patients with AVMs were managed conservatively and 103 patients underwent intervention. Due to the observational nature of the study, patients who received intervention were younger, more likely to present with seizure, more likely to have angiogram, and less likely to have maximum AVM diameter greater than 6 cm. Sixty-eight patients underwent single treatment: 28 radiosurgeries, 22 en-dovascular embolizations, and 18 microsurgeries. Thirty-five patients underwent multimodality treatment: 20 embolization and radiosurgery, 12 embolization and microsurgery, 2 radiosurgery and embolization, and 1 had all the 3 treatments. Overall AVM obliteration rate was 66%.

Median follow-up was 6.8 years (IQR 4-11 y), accounting for a total of 1,479 person-years. For the first 4 years, the risk of primary outcome was greater in the intervention group than the conservatively managed group (HR 0.59, 95% CI 0.35-0.99). However, in the subsequent time periods, 4 to 8 and 8 to 12 years, no significant difference in primary outcome was observed. The overall 12-year death rate was higher in the conservatively managed group (31 vs. 10 deaths); however, this was attributable to deaths from other causes and statistical significance disappeared after adjusting for age. Due to a large incidence of symptomatic stroke following intervention, the risk for progression to secondary outcome was significantly lower in the conservatively managed patients over the entirety of the study (HR 0.37, 95% CI 0.19-0.72).

The SIVMS investigation was an observational study that ultimately had comparable results to those of the ARUBA trial. While the SIVMS study circumvented many of the critiques that ARUBA received, it was not without limitations. Due to the observational nature of the study, patients treated with interventions were naturally younger, more likely to present with seizure, and had smaller AVM nidal diameter. Furthermore, since the intervention cohort was younger, more patients suffered non-AVM death in the conservatively managed arm, rendering the primary outcome difficult to interpret beyond the first 4 years of the study. Like ARUBA, critics argue that further follow-up of the SIVMS cohort is warranted to assess if the reduced risk associated with conservative management persists.

More than ever, the ARUBA and SIVMS studies call upon our code of “first do no harm.” While both studies have significant limitations and flaws, they investigate an important topic that still required further elucidation. One must use their clinical judgment and critical understanding of the published literature, in addition to specific patient and AVM characteristics when deciding to intervene or conservatively manage unruptured AVMs. As such, many centers have further refined their patient selection criteria especially for high-risk patients leading to decreased morbidity and mortality.