Brain arteriovenous malformations (BAVMs) consist of a tangle of arterialized venular vessels without interposing capillaries, presenting with spirally coiled shapes, which form numerous fenestrated channels; they intercommunicate and empty into thin-walled draining veins (1). Endovascular treatment (EVT) can be used for presurgical or preradiosurgical treatment of BAVMs or as a stand-alone procedure for curative purposes (2, 3). The main role of EVT is either to diminish the nidus size or to occlude high-risk features, such as ruptured nidal and perinidal aneurysms (4).

Currently, the use of the Onyx liquid embolic system (Medtronic, Irvine, CA, USA) is popular in EVT for BAVMs (5). However, EVT with Onyx embolization for BAVMs is not always safe and can be associated with many complications, among which intra- or post-operative acute hemorrhage is disastrous (6, 7). Currently, more summaries of experiences from different centers are required to achieve a full understanding of this type of complication. Therefore, this retrospective study investigated this issue in 25 consecutive patients who experienced this type of complication.

During the period from January 2010 to December 2021, 673 patients diagnosed with a BAVM accepted EVT with 856 Onyx embolization procedures. Of them, 25 experienced intra- or post-operative acute hemorrhage, for a rate of 3.7% (25/673) per patient and 2.9% (25/856) per procedure.

Except for SM grade IV or V BAVMs, which are generally monitored by observation unless ruptured, EVT plays an important role in securing high-risk bleeding points (such as flow-related or intranidal aneurysms and fistulous components) or reducing the flow to help shrink the nidus or area of perinidal angiogenesis (4, 26). For SM grade I or II BAVMs, EVT can be used as a curative treatment, resulting in total occlusion of the nidus and filling of the proximal part of the draining vein (27–30). However, curative embolization of BAVMs should be considered an unanticipated benefit of such therapy rather than a goal (31). For SM grade III BAVMs, EVT can reduce the BAVM volume before radiosurgery and address high-risk bleeding points (32–34). When BAVMs have large or deep feeding arteries, pre-operative embolization can be used to reduce the flow greatly, prevent intraoperative bleeding, and reduce local venous hypertension and vascular steal, which facilitates microsurgical resection (35).

However, EVT for BAVMs is a challenge, especially considering hemorrhagic complications, which are the most frequent complication of EVT, reported in 2–12.5% of procedures (6, 36). In a study by Baharvahdat et al., including 846 embolization procedures in 408 patients, hemorrhagic complications occurred in 11% of procedures (6). In a report by Sato et al. with a total of 1042 EVT procedures, the rate of hemorrhagic complications was 5.7% (36). In a report by Liu et al. describing the use of Onyx as an embolic agent in 126 patients with 143 consecutive interventions, the periprocedural bleeding complication rate was 5.4% per patient and 4.7% per procedure (37).

In our study, the rate was 3.4% per patient and 2.9% per procedure, which is lower than that in the above reports. The main reason for this difference is that our study excluded hemorrhagic complications occurring more than 72 h post-operatively, which resulted in the inclusion of fewer cases. At our center, microwire navigation and microcatheter catheterization are performed very cautiously. Therefore, the rate of arterial rupture due to technical complications is not as high as that reported by Baharvahdat et al. in their report, 48% of hemorrhagic complications were related to arterial rupture (6); in our study, the corresponding rate was 16%. In addition, in our study, the pressure cooker technique was applied in 12.1% of procedures. After Onyx casting, due to the lack of a long reflex distance of Onyx, the coiling did not fix the microcatheter, and the microcatheter could be easily retrieved without difficulty, as suggested in our previous reports (38, 39).

High-pressure Onyx casting plays an important role in the intraoperative rupture of BAVMs, and this type of complication should also be considered a technical complication (16). During Onyx casting, due to intercompartmental communications, theoretically, a single Onyx casting may obliterate all compartments of the nidus. However, when continuing the casting to push excess Onyx from one compartment into other compartments, the increased pressure of Onyx casting can lead to rupture of the BAVM structure, especially in cases with intranidal aneurysms or even feeding artery aneurysms (16, 18, 40). In our study, rupture from high-pressure Onyx casting accounted for 76.5% of intraoperative hemorrhages. During Onyx casting, early primary draining vein occlusion will increase the resistance to Onyx casting and increase the risk of intraoperative hemorrhage (41). Among 17 intraoperative hemorrhagic complications, there were four perforating complications; of the remaining 13 complications, there were four primary draining vein occlusions, accounting for 30.8% (4/13). Therefore, primary draining vein occlusion should be avoided.

When a BAVM exhibits a monocompartmental compact nidus with one feeder, elimination of the single feeder leads to the collapse of the entire BAVM (13). However, in our study, with 25 BAVMs, 56% of BAVMs had secondary feeding arteries, and 40% of BAVMs had secondary draining veins, which indicated that many BAVMs were multicompartmental, resulting in increased resistance to Onyx casting and a greater risk of intraoperative hemorrhage (16). Therefore, when there is resistance, EVT should be performed with caution. To reduce the occurrence of intraoperative hemorrhagic complications, Cai et al. proposed simultaneous Onyx casting via dual catheters because injection from two different feeders can result in the gradual, centripetal exclusion of the nidus (42). However, more evidence of the technique is required to support its application in clinical practice.

Currently, definite causes of post-operative hemorrhagic complications are unclear. Only some hypotheses have been proposed. According to Baharvahdat et al. post-operative hemorrhagic complications may originate from normal perfusion pressure breakthrough that can result in the rupture of a non-embolized nidus or area of perinidal angiogenesis due to increased hemodynamic stress in this region. Complete or incomplete draining vein occlusion can also result in post-operative rupture of the nidus or area of perinidal angiogenesis due to venous hyperemia or progressive flow slowing or thrombosis of the draining vein. In addition, inflammatory reactions or mural necrosis induced by the embolic material can result in post-operative hemorrhage (6).

In our study, except for two (25%, 2/8) hemorrhages attributed to draining vein system rupture, six (75%, 6/8) post-operative hemorrhages were attributed to normal perfusion pressure breakthrough. The casting of an excessive amount of Onyx at one time can increase the occurrence of post-operative normal perfusion pressure breakthrough (43–47). In a report by Katsaridis et al. 5% (5/101) of patients had post-operative hemorrhagic complications; in these patients, the degree of nidus embolization with Onyx was often more than 80% (48). In our study, among eight post-operative hemorrhages, the degree of nidus embolization with Onyx was more than 2/3 of the nidus or the complete nidus in 87.5% of cases. Therefore, complete embolization of the nidus should not be pursued in EVT for BAVMs (27, 30, 31). Staged embolization with Onyx over 4–6 months can be helpful, with < 30% obliteration of the nidus every time (7, 26).

In our study, of the eight BAVMs with post-operative complications, primary draining vein occlusion occurred in four, accounting for 50%, which increased the risk of post-operative hemorrhagic complications. When the draining function of the primary draining vein is impaired due to unexpected Onyx occlusion or curative embolization, venous congestion in the adjacent brain region might progress through delayed thrombosis in the draining vein, and the scant venous flow may result in delayed occlusion of the remaining drainers, resulting in post-operative venous hemorrhage (3, 49). Our study included two confirmed draining vein system ruptures. Therefore, the procedure of EVT for BAVMs should not compromise venous drainage, especially the plexiform component or dural sinuses of the drainage vein system (50).

Intra- and post-operative acute hemorrhagic complications may be disastrous; in our study, 40% of patients died. For intraoperative hemorrhage, the only treatment available is continuous embolization until the bleeding stops. If the rupture point is small, there is less bleeding, and a good prognosis can be achieved; otherwise, the bleeding can be disastrous or fatal. EVT cannot be performed for post-operative hemorrhage, and the prognosis is determined by the severity and location of the hemorrhage.

For intraoperative hemorrhagic complications, in addition to vessel rupture-related complications, high-pressure Onyx casting is an important risk factor. For post-operative acute hemorrhagic complications, casting an excessive amount of Onyx at one time to increase the degree of nidus embolization may result in post-operative hemorrhage due to normal perfusion pressure breakthrough. Primary draining vein occlusion can also increase the risk of intra- and post-operative acute hemorrhage. Hemorrhagic complications can be disastrous; in our study, only 60% of patients had a good outcome.

This was a retrospective study without a control group. The conclusion should be interpreted with caution. In addition, the results cannot indicate whether the rupture status of BAVMs prior to EVT is related to the occurrence of post-operative hemorrhagic complications, whether hemorrhage complications tend to occur in infratentorial BAVMs or high SM grade BAVMs, or whether the pressure cooker technique increases hemorrhage complications.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

Written informed consent was obtained from the individual(s), and minor(s)' legal guardian/next of kin, for the publication of any potentially identifiable images or data included in this article.

JY and XC contributed to the conception, design of the manuscript, and critically revised the manuscript. XC and YW wrote the manuscript and collected the medical records of the patients. All authors approved the final version of this manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.