Intracranial dural arteriovenous fistula (DAVF) is an acquired arteriovenous shunt between the dural arteries and cortical veins or dural sinuses [1,2]. Since Fincher’s initial report in 1951 [3], DAVF cases have continuously been documented. DAVFs are now recognized as one of the main forms of central nervous system vascular malformations. The associated symptoms include bruit and venous impairments secondary to intracranial venous occlusion, stasis, and reflux. These abnormalities include intracranial hemorrhage, ocular symptoms, and non-hemorrhagic neurological deficits (NHNDs) [1,2]. As suggested by commonly used grading systems, such as those of Borden et al. [4] and Cognard et al. [5], the risk posed by DAVF is mainly dependent on the presence of cortical venous reflux (CVR [6,7]). DAVFs without CVR rarely cause hemorrhage or NHND. DAVFs with CVR have an annual risk of hemorrhage of up to 7.6% [8]. Nevertheless, low-risk DAVFs may also develop similar aggressive features over time [9,10].

Stereotactic radiosurgery (SRS), which involves the delivery of a precise, high-dose of radiotherapy in a single session, is known to be effective for treating arteriovenous shunt disorders [11]. The relevant evidence has been obtained mainly from studies on arteriovenous malformations (AVMs). However, there remains a paucity of high-quality evidence from largescale investigations of patients with DAVFs. Accordingly, many neurosurgeons consider endovascular embolization and direct surgery as the current standard therapeutic modalities for DAVFs. SRS is generally considered an alternative when endovascular embolization and/or direct surgery fail or are not feasible. Nevertheless, a gradual but steady accumulation of retrospective evidence suggests that SRS has a similar level of efficacy to other modalities. In addition, SRS has the added advantage of minimal invasiveness.

Several issues must be clarified before establishing SRS as an effective treatment option for DAVFs. First, the factors associated with favorable outcomes must be clarified. The creation of a practical radiosurgical grading scale would be invaluable for guiding the optimal application of SRS. Second, many patients with DAVF undergo SRS with embolization (Emb-SRS). Evaluating SRS outcomes as a standalone treatment and comparing them to Emb-SRS treatment is essential to develop a better understand of the role of SRS as a standalone modality. To accomplish these goals, a study that includes many patients is required for appropriate statistical analyses. However, such studies are challenging to perform at a single institution, given the rarity of SRS cases. The present nationwide, multi-institutional study was conducted in collaboration with the Japanese Leksell Gamma Knife Society.

In this retrospective, multi-institutional study on SRS for DAVF, the overall outcomes were satisfactory, with a favorable obliteration rate (78.8% at 4 years) and a low symptomatic ARE rate (1.0%). These results are largely in line with those of earlier studies [16-26]. Approximately 75% of post-SRS hemorrhages occurred during the latency phase during the process of fistula obliteration. Although the lack of an immediate effect is a major shortcoming of SRS, the hemorrhage rate (2.0%) was acceptable, and the neurological sequelae were generally mild. Improvements in symptoms, especially bruit and ophthalmic symptoms, were achieved in most patients and occurred earlier than fistula obliteration (median, 5.0 to 6.0 months). The number of patients (n=200) in our study is the largest ever reported, which supports the validity and robustness of our results. The main achievement of this study was the development of a scoring system. This system was based on the location category and the presence/absence of CVR that could successfully predict the likelihood of SRS-induced fistula obliteration. The area under the curve of the system (0.75) indicated a good predictive accuracy [27], and the R² value (0.24) was comparable to that achieved using the Virginia Radiosurgery AVM Scale (0.090) [15]. No grading system predicting the results of SRS for DAVF has ever been established. Consequently, our scoring system appears to be especially practical and may guide surgeons in determining more appropriate, situation-based therapeutic strategies. The generalizability of this finding requires further evaluation in future studies.

Identifying the factors associated with favorable outcomes is essential for the safe and effective application of SRS. As noted in a previous study, the commonly used grading scales for DAVFs, such as Borden’s and Cognard’s scales, are not useful in this regard [28]. Because of the rarity of both DAVF itself and its treatment with SRS, few previous studies involving relatively large numbers of patients have found factors possibly associated with fistula obliteration. Examples of these factors include absence of CVR [16,19], CS location [16,28], female sex [28], and absence of venous ectasia [28]; however, no factor has been consistently identified. Our results confirmed a significant association between the absence of CVR and obliteration. The reason for this association remains unclear; however, hemodynamic factors may play a significant role, given that the presence of CVR implies intolerably increased pressure in the shunting point/draining sinus. Another factor may be the large diameters of the shunting vessels, which presumably lead to a high-flow shunt and thus an increase in pressure. Regarding fistula location, we not only confirmed the previous finding that CS location was associated with obliteration but also observed that the likelihood of obliteration could be stratified by the location category designated as lowest (ASB and MF), intermediate (SSS and Tent), and highest (CS, TS, and others). DAVFs in CS/TS are in the proximity of the large dural sinuses, while those in the ASB/MF are not. One possible explanation for this association is that an anatomical relationship to a large dural sinus can potentially “buffer” elevated venous pressure. In contrast, although female sex was significantly associated with obliteration in bivariate analysis, an independent association was not observed in the multivariate analysis. The strong female predominance of CS-DAVFs [28-30] seems to explain this ostensible relationship. Indeed, a significant association between female sex and obliteration has not been confirmed in most recent studies [16,19,30-32]. We did not collect information on venous ectasia. Thus, the significance of this factor remains to be elucidated. However, venous ectasia theoretically occurs because of elevated venous pressure in the cortical veins with regurgitation, which seems to have the same pathophysiological significance as CVR. Unlike AVMs, radiosurgical doses or target volumes were not significantly associated with obliteration, similar to previous studies [16,19,30-32]. This result may be attributed to the fact that these factors were more homogeneous than AVM characteristics. Despite our results, prescription doses should be maintained at 18 Gy or higher, given that such doses have been used in many previous studies as well as the present study [16,28,32-36].

A combination of SRS and embolization has been documented in several studies and has become a commonly used treatment strategy for DAVFs [25,26,36]. Although the true effect of SRS on AVMs remains debatable, several studies have suggested that embolization negatively affects SRS outcomes [16,37]. In contrast, embolization can be safely incorporated into SRS for DAVFs. Our comparisons and matched cohort analyses showed no negative effects of embolization. This finding also implies that embolization combined with SRS may not confer a definitive benefit. However, we should interpret these results with caution, as our cohort was likely subject to selection bias. In fact, NHNDs were significantly less common in the SRS-alone group than in the Emb-SRS group. Nevertheless, in the SRS alone group, there were comparable numbers of patients with purported aggressive features such as CVR (38.6% vs. 48.7%), non-sinus-type DAVFs (13.3% vs. 20.0%), and hemorrhagic onset (16.9% vs. 21.4%). The latter of these two factors did not appear to significantly reduce the obliteration rate. Although CVR has been demonstrated to reduce the obliteration rate, fistula obliteration could be achieved in approximately 70% of these patients. Given the satisfactorily low rates of symptomatic AREs and post-SRS hemorrhages, SRS alone appears to be safe and effective and may be worth considering for such patients. However, it should be noted that endovascular embolization and subsequent early obliteration or flow reduction of the arteriovenous shunt may be beneficial for impending cases, such as DAVFs with early rebleeding and progressive focal neurological deficits other than ophthalmoplegia.

This study has a few limitations. First, selection bias was unavoidable given the retrospective nature of the study. Therefore, cautious interpretation of the results of comparisons between SRS alone and Emb-SRS is necessary, as described above. This fact might also impair the external validity of the new scoring system, for which further evaluation is required. Therefore, the generalizability of this study to individual cases should be considered. Moreover, multi-institutional studies tend to be subject to reporting bias. Therefore, to minimize bias, complete enumeration was mandated in this study. Furthermore, fistula obliteration was confirmed based on magnetic resonance imaging/computed tomography angiography findings in 19.3% of patients; thus, a slight overestimation of the obliteration rates might have occurred. Finally, although Emb-SRS exhibited the same level of efficacy as SRS alone, embolization-associated adverse events must be considered.

SRS was found to be effective for DAVFs, with an estimated 4-year cumulative obliteration rate of 78.8%, annual post-SRS hemorrhage rates of 0.8% (≤2 years) and 0.5% (>2 years), and a crude symptomatic ARE rate of 1.0%. In addition to the fact that embolization does not appear to adversely affect SRS, SRS alone may have a similar level of efficacy to Emb-SRS and can be considered feasible, except in impending situations in which endovascular embolization and subsequent early obliteration or flow reduction of the arteriovenous shunt may be beneficial. It remains unclear whether Emb-SRS is superior to SRS alone in certain DAVFs. Multivariate analysis revealed that CVR and location category (ASB/MF vs. SSS/Tent vs. CS/TS/others) were associated with fistula obliteration. The radiosurgical grading system developed using these factors could successfully stratify patients into three groups: highest, intermediate, and lowest expectations of obliteration. These findings may help determine more appropriate and tailored therapeutic strategies for DAVFs.