Clinical Outcomes of Fractionated Radiation Therapy for Spinal Intramedullary Arteriovenous Malformations: A Single Institutional Retrospective Review Authors: Ruja Vichitvejpaisal, Kentaro Nishioka, Toshiya Osanai, Motoyuki Iwasaki, Takashi Mori, Yusuke Uchinami, Tomohiko Miyazaki, Fuki Koizumi, Takayuki Hashimoto, Miki Fujimura, Hidefumi Aoyama Published: 2026 ABSTRACT PurposeIntramedullary spinal arteriovenous malformations (im-sAVMs) have a bleeding risk of up to 4% to 10% annually. Given the rarity of im-sAVMs and the lack of established evidence regarding the efficacy of radiation therapy, we retrospectively reviewed the treatment outcomes of fractionated radiation therapy (fRT) administered for im-sAVMs at our institution.Methods and MaterialsWe examined patients of im-sAVMs treated with fRT at Hokkaido University Hospital between January 2003 and December 2023. The incidences of post-fRT bleeding from the AVMs, nidus occlusion/reduction, and late radiation-induced complications were assessed.ResultsOf the 25 patients, we analyzed 21 patients, excluding 4 patients with a follow-up period of less than 180 days. Prior to radiation therapy, 10 patients underwent surgical occlusion of feeding vessels because of coexisting arteriovenous fistulas, 2 and 3 patients underwent surgical occlusion and endovascular embolization for arteriovenous malformation, respectively. The radiation dose was 20 Gy in 4 fractions for 19 patients and 24 Gy in 4 fractions for 2 patients. The follow-up period ranged from 12 to 236 months (median, 56.4 months). Complete nidus occlusion was observed in 2 patients (9.5%), whereas 8 patients (38.1%) showed partial reduction with residual nidus, and 10 patients (47.6%) showed no reduction. One patient showed slight nidus enlargement 12 months after fRT, and another patient developed an aneurysm formation in the feeding artery at the margin of the irradiation field 57 months after fRT. No bleeding and late radiation-induced complications were observed during the follow-up.ConclusionsFractionated radiation therapy at doses of 20 to 24 Gy in 4 fractions for im-sAVMs was shown to be safely administered and may be effective in reducing the risk of bleeding. Introduction Spinal arteriovenous malformations (sAVMs) are rare vascular malformation with an unknown cause and natural histories. The symptoms include bleeding (4% per year and can increase to 10% per year after first bleeding1), compressive myelopathy, venous hypertension, and steal phenomenon. This type of vascular abnormality could cause an inferior prognosis if left untreated. The patients might suffer from acute, subacute, and chronic spinal cord dysfunction, leading to severe permanent neurologic deterioration such as myelopathy and radiculopathy.2,3 Generally, there are 4 types of sAVMs, classified based on the location of the arteriovenous shunt,4,5 as described below (Fig. 1):A)Dural arteriovenous fistula (AVF): Most common in middle-aged male patients, this type of shunt is located on the dura surface near the intervertebral foramen and refluxes backward into the vein on the surface of the spinal cord. Symptoms usually arise from increased venous pressure, leading to progressive flaccid paraplegia.B)Epidural AVF: In this type, the arteriovenous shunt is located outside the dura but does not reflux inward as in the dural AVF type. Instead, the shunt flows directly into the epidural vein. This type is usually asymptomatic but can cause symptoms because of nerve root compression.C)Perimedullary AVF: The arteriovenous shunt is located on the surface of the spinal cord. If the shunt is positioned at the cervical level, subarachnoid or intramedullary hemorrhage often occurs. In contrast, symptoms from increased venous pressure are more common when the shunt is located at the thoracic level.D)Intramedullary sAVMs (im-sAVMs): Primarily found in younger patients, this type of arteriovenous malformation involves the shunt or nidus being located within the spinal cord, which can cause intramedullary hemorrhage.Figure 1Classification of spinal arteriovenous malformations based on the location of the arteriovenous shunt (A) dural arteriovenous fistula (AVF); (B) epidural AVF; (C) perimedullary AVF; and (D) intramedullary arteriovenous malformation.Figure 1 dummy alt text A)Dural arteriovenous fistula (AVF): Most common in middle-aged male patients, this type of shunt is located on the dura surface near the intervertebral foramen and refluxes backward into the vein on the surface of the spinal cord. Symptoms usually arise from increased venous pressure, leading to progressive flaccid paraplegia. Dural arteriovenous fistula (AVF): Most common in middle-aged male patients, this type of shunt is located on the dura surface near the intervertebral foramen and refluxes backward into the vein on the surface of the spinal cord. Symptoms usually arise from increased venous pressure, leading to progressive flaccid paraplegia. B)Epidural AVF: In this type, the arteriovenous shunt is located outside the dura but does not reflux inward as in the dural AVF type. Instead, the shunt flows directly into the epidural vein. This type is usually asymptomatic but can cause symptoms because of nerve root compression. Epidural AVF: In this type, the arteriovenous shunt is located outside the dura but does not reflux inward as in the dural AVF type. Instead, the shunt flows directly into the epidural vein. This type is usually asymptomatic but can cause symptoms because of nerve root compression. C)Perimedullary AVF: The arteriovenous shunt is located on the surface of the spinal cord. If the shunt is positioned at the cervical level, subarachnoid or intramedullary hemorrhage often occurs. In contrast, symptoms from increased venous pressure are more common when the shunt is located at the thoracic level. Perimedullary AVF: The arteriovenous shunt is located on the surface of the spinal cord. If the shunt is positioned at the cervical level, subarachnoid or intramedullary hemorrhage often occurs. In contrast, symptoms from increased venous pressure are more common when the shunt is located at the thoracic level. D)Intramedullary sAVMs (im-sAVMs): Primarily found in younger patients, this type of arteriovenous malformation involves the shunt or nidus being located within the spinal cord, which can cause intramedullary hemorrhage. Intramedullary sAVMs (im-sAVMs): Primarily found in younger patients, this type of arteriovenous malformation involves the shunt or nidus being located within the spinal cord, which can cause intramedullary hemorrhage. Classification of spinal arteriovenous malformations based on the location of the arteriovenous shunt (A) dural arteriovenous fistula (AVF); (B) epidural AVF; (C) perimedullary AVF; and (D) intramedullary arteriovenous malformation. The mainstay of treatment for sAVMs types A to C is microsurgery and/or endovascular embolization. However, these treatment options are not suitable for every patient, particularly those with im-sAVMs, because of the risk of neurologic deficits, such as spinal infarction. Boström et al6 reported a 20% rate of neurologic worsening following microsurgery, with 5% of cases resulting in permanent neurologic deterioration. Similarly, Velat et al7 reported a surgical morbidity rate of 5% after microsurgery. As a result, fractionated radiation therapy (fRT) is considered an excellent noninvasive alternative for treating im-sAVMs when microsurgery or endovascular embolization is not feasible. The radiation therapy technique has advanced over the past few decades, enabling radiation oncologists to treat patients more accurately and effectively. At Hokkaido University Hospital, we have treated patients with im-sAVMs for over a decade. This retrospective review was conducted at our institution to evaluate whether fRT is beneficial in treating im-sAVMs without causing long-term, permanent radiation toxicity. Methods and Materials This single-center, retrospective study was approved by the institutional review board. The materials and methods are as follows. Patient selection The paper and electronic medical records of patients with im-sAVMs treated with fRT at Hokkaido University Hospital between January 2003 and December 2023 were reviewed. Patients with a follow-up period of less than 180 days were excluded from the study. Evaluation and treatment All patients were evaluated using angiography and magnetic resonance imaging (MRI) before radiation therapy. CT simulation was performed for treatment planning. Linac-based fRT was used, with doses ranging from 20 to 24 Gy delivered in 4 fractions to the nidus area. Treatment response was evaluated using either MRI or angiography. Study endpoints The endpoints of this study were the incidence of bleeding after fRT, the obliteration rate of the nidus, and late radiation-induced toxicity. Basic patient characteristics, treatment planning procedures, and all study endpoints were analyzed using descriptive statistics, presented as means, medians, and percentages. Results Among 25 patients with im-sAVMs who underwent fRT between January 2003 and December 2023, 4 patients were excluded because of follow-up times of less than 180 days, leaving 21 patients enrolled in the study. The median age at the time of radiation therapy was 27 years (range, 14-55); 10 were male and 11 were female biologically. Of the 21 patients, the distribution of sAVMs included 9 at the cervical level, 10 at the thoracic level, and 2 at the lumbar level. Hemorrhage was present at initial presentation in 42.8% of patients, and 71.4% (15 of 21) had undergone prior treatment of coexisting AVF or arteriovenous malformations (AVMs) with microsurgery or endovascular embolization before radiation therapy (10 with microsurgery for AVF, 2 with microsurgery and 3 with endovascular embolization for AVMs). Only 4.8% (1 of 21) underwent microsurgery for the residual coexisting AVF after fRT, while no patients received treatment for AVM. The details of the patients’ characteristics are shown in Table 1 and supplementary material (Table E1).Table 1Patient characteristics (n = 21)Table 1 dummy alt textCharacteristicValueRatio (%)AgesMedian (y)27 (14-55)-Follow-up time (mo)- Median56.4 (12-236)- <36523.8 37-72942.9 73-108419.0 >108314.3Biological gender Male1047.6 Female1152.4Distribution Cervical942.9 Thoracic1047.6 Lumbar29.5Prior treatment of coexisting AVF Microsurgery1047.6 Embolization00 None1152.4Prior treatment of AVM Microsurgery29.5 Embolization314.3 None1676.2Treatment after fRT Microsurgery for residual AVF14.8 Embolization00 None2095.2Immobilization Shell/gold marker314.3 Shell628.6 Cushion29.5 Gold marker628.6 None419.0Radiation therapy technique 3DCRT1990.5 IMRT29.5Radiation therapy dose 20 Gy/4 F1990.5 24 Gy/4 F29.5Abbreviations: 3DCRT = 3-dimensional conformal radiation therapy; AVF = arteriovenous fistula; AVM = arteriovenous malformation; fRT = fractionated radiation therapy; IMRT = intensity modulated radiation therapy. Table 1Patient characteristics (n = 21)Table 1 dummy alt textCharacteristicValueRatio (%)AgesMedian (y)27 (14-55)-Follow-up time (mo)- Median56.4 (12-236)- <36523.8 37-72942.9 73-108419.0 >108314.3Biological gender Male1047.6 Female1152.4Distribution Cervical942.9 Thoracic1047.6 Lumbar29.5Prior treatment of coexisting AVF Microsurgery1047.6 Embolization00 None1152.4Prior treatment of AVM Microsurgery29.5 Embolization314.3 None1676.2Treatment after fRT Microsurgery for residual AVF14.8 Embolization00 None2095.2Immobilization Shell/gold marker314.3 Shell628.6 Cushion29.5 Gold marker628.6 None419.0Radiation therapy technique 3DCRT1990.5 IMRT29.5Radiation therapy dose 20 Gy/4 F1990.5 24 Gy/4 F29.5Abbreviations: 3DCRT = 3-dimensional conformal radiation therapy; AVF = arteriovenous fistula; AVM = arteriovenous malformation; fRT = fractionated radiation therapy; IMRT = intensity modulated radiation therapy. Patient characteristics (n = 21) Abbreviations: 3DCRT = 3-dimensional conformal radiation therapy; AVF = arteriovenous fistula; AVM = arteriovenous malformation; fRT = fractionated radiation therapy; IMRT = intensity modulated radiation therapy. All patients underwent angiography and MRI before radiation therapy to locate the AVM nidus. CT simulation was done for radiation therapy planning. Among the cases that underwent surgical treatment prior to fRT, fiducial markers were placed in the vertebrae near the nidus in 9 patients and were used for patient setup. Patient immobilization techniques are shown in Table 1. In the first 19 cases, we used the 3-dimensional conformal radiation therapy (3DCRT) technique, whereas the intensity modulated radiation therapy (IMRT) technique was used for planning in the last 2 cases. For 3DCRT planning, the clinical target volume included the entire spinal canal, covering all levels of the nidus region. In the last 2 patients where IMRT was used, the clinical target volume included only the AVM nidus, which was delineated by an experienced neuro-oncologist using MRI combined with angiography to accurately localize the nidus (Fig. 2). The nidus volume was calculated in 2 patients treated with the IMRT technique, showing volumes of 1.7 cc and 2.3 cc, respectively. However, it was not possible to access the treatment planning system for the patients treated with 3DCRT, and therefore, volume data for the nidus could not be collected. The prescribed dose was 20 Gy in 4 fractions for 19 patients and 24 Gy in 4 fractions for 2 patients. In cases treated with 3DCRT, the dose was prescribed to the isocenter set at the center of the nidus. For cases treated with IMRT, the prescribed dose was set to ensure that over 95% of the gross tumor volume (GTV, ie, nidus) was covered by the prescribed dose and that the minimum dose to the planning target volume (PTV, ie, GTV + 1 mm) reached at least 90% of the prescribed dose.Figure 2(Patient No. 20) A 26-year-old male presented with paralysis of both lower limbs and later developed weakness in the upper limbs as well as the neurogenic bladder. (A) (Left) Magnetic resonance imaging (MRI) showed flow voids on T2-weighted images at the C6 to C7 level (white arrow) with intramedullary hemorrhage extending from C5 to C7. (right) Digital subtraction angiography (DSA) revealed intramedullary spinal arteriovenous malformation with a nidus measuring approximately 2.8 × 1.4 cm at the C6 to C7 level. The feeder vessel originated from the left T4 intercostal artery, flowing via the anterior spinal artery. The patient underwent fractionated radiation therapy (fRT) at a dose of 24 Gy in 4 fractions using the intensity modulated radiation therapy technique. (B) Shows the gross tumor volume (GTV) (ie, nidus, in red) and isodose distribution. The 100% isodose line (navy) covered at least 95% of the GTV, ensuring that the entire planning target volume (GTV + 1mm) was included within the 90% isodose line (green). (C) (Left) Shows MRI and (right) DSA at 4.8 months after fRT. The nidus (white arrow) was dramatically decreased in size. (D) (Left) shows MRI at 39.7 months after fRT where all flow voids had disappeared. (right) The DSA confirmed the disappearance of the nidus compared with before fRT. Markedly high T2 signal was noted at the C6 to C7 level (yellow arrow), indicating fatty marrow changes from radiation therapy.Figure 2 dummy alt text (Patient No. 20) A 26-year-old male presented with paralysis of both lower limbs and later developed weakness in the upper limbs as well as the neurogenic bladder. (A) (Left) Magnetic resonance imaging (MRI) showed flow voids on T2-weighted images at the C6 to C7 level (white arrow) with intramedullary hemorrhage extending from C5 to C7. (right) Digital subtraction angiography (DSA) revealed intramedullary spinal arteriovenous malformation with a nidus measuring approximately 2.8 × 1.4 cm at the C6 to C7 level. The feeder vessel originated from the left T4 intercostal artery, flowing via the anterior spinal artery. The patient underwent fractionated radiation therapy (fRT) at a dose of 24 Gy in 4 fractions using the intensity modulated radiation therapy technique. (B) Shows the gross tumor volume (GTV) (ie, nidus, in red) and isodose distribution. The 100% isodose line (navy) covered at least 95% of the GTV, ensuring that the entire planning target volume (GTV + 1mm) was included within the 90% isodose line (green). (C) (Left) Shows MRI and (right) DSA at 4.8 months after fRT. The nidus (white arrow) was dramatically decreased in size. (D) (Left) shows MRI at 39.7 months after fRT where all flow voids had disappeared. (right) The DSA confirmed the disappearance of the nidus compared with before fRT. Markedly high T2 signal was noted at the C6 to C7 level (yellow arrow), indicating fatty marrow changes from radiation therapy. During the follow-up period, clinical and radiographic evaluations (either angiography or MRI) were performed for all patients at each follow-up visit. At a median follow-up time of 56.4 months (range, 12-236), the AVM nidus was obliterated in 9.5% of patients (2 of 21), reduced in size in 38.1% (8 of 21), remained stable in 47.6% (10 of 21), and increased in size in 4.8% (1 of 21). The median time from the first day of radiation therapy to a reduction in size was 14 months (range, 4.8-37.5). Of the 2 patients whose AVM was completely obliterated, one occurred at 39.7 months after radiation therapy and the other at 52 months. All details regarding the nidus response after radiation therapy, including additional treatments, are shown in the swimmer plot (Fig. 3). No patients experienced hemorrhagic symptoms or proven hemorrhage by radiographic evaluation after fRT until the last follow-up. No radiation therapy-induced late adverse event was observed. However, one patient developed an aneurysm in the feeding artery 57 months after radiation therapy. Because the aneurysm occurred outside the radiation field, it is unlikely to have been caused by the radiation (Fig. 4).Figure 3Swimmer plot showing nidus response and additional treatments after radiation therapy in relation to the duration of follow-up.Figure 3 dummy alt textFigure 4(Patient No. 5) A 31-year-old female presented with weakness in the left lower limb, and angiography revealed intramedullary spinal arteriovenous malformation at the T3 to T4 level. After undergoing surgical treatment for a perimedullary arteriovenous fistula, she received fractionated radiation therapy (fRT) at a dose of 20 Gy in 4 fractions. Flow voids had decreased in size by 20.9 months after fRT, with no bleeding detected. However, magnetic resonance imaging at 57 months post-fRT (left) showed a new aneurysm at the T2 level (white arrow). A review of the previous fRT planning (right) confirmed that the aneurysm was located above the irradiated area.Figure 4 dummy alt text Swimmer plot showing nidus response and additional treatments after radiation therapy in relation to the duration of follow-up. (Patient No. 5) A 31-year-old female presented with weakness in the left lower limb, and angiography revealed intramedullary spinal arteriovenous malformation at the T3 to T4 level. After undergoing surgical treatment for a perimedullary arteriovenous fistula, she received fractionated radiation therapy (fRT) at a dose of 20 Gy in 4 fractions. Flow voids had decreased in size by 20.9 months after fRT, with no bleeding detected. However, magnetic resonance imaging at 57 months post-fRT (left) showed a new aneurysm at the T2 level (white arrow). A review of the previous fRT planning (right) confirmed that the aneurysm was located above the irradiated area. Case study 1 A 26-year-old male presented with paralysis of both lower limbs and later developed weakness in the upper limbs as well as the neurogenic bladder (patient No. 20, Fig. 2). MRI revealed flow voids on T2-weighted images at the C6 to C7 level with intramedullary hemorrhage extending from C5 to C7. Digital subtraction angiography (DSA) showed im-sAVMs with a nidus measuring approximately 2.8 × 1.4 cm at the C6 to C7 level. The feeder vessel originated from the left T4 intercostal artery, flowing via the anterior spinal artery (ASA). Following conservative treatment, the patient's weakness improved, and he was referred to Hokkaido University Hospital for definitive treatment. Because of the high surgical and embolization risks associated with ASA involvement, we decided to use radiation therapy as the primary treatment approach. The patient underwent fRT at a dose of 24 Gy in 4 fractions over consecutive days using the IMRT technique. The GTV included the nidus, which was delineated by combining CT angiography with CT simulation images and MRI. The PTV consisted of the GTV plus a 1 mm margin to account for setup error. The prescribed dose was designed to cover at least 95% of the GTV, ensuring that the entire PTV was within the 90% isodose line. At 4.8 months posttreatment, both MRI and DSA showed a dramatic reduction in nidus size. At the 39.7-month follow-up, the MRI revealed complete nidus resolution, which was confirmed by DSA. His symptoms continued to improve over time, with only mild left cervical radiculopathy at the C7 level, and he was able to resume daily life as usual. Case study 2 A 31-year-old female presented with weakness in the left lower limb (patient No.5, Fig. 4). She had previously undergone 3 endovascular embolizations for a dural AVF at another hospital. After 12 years of follow-up, her weakness worsened, and she developed a neurogenic bladder, requiring self-catheterization. She was subsequently referred to Hokkaido University Hospital. Angiography revealed im-sAVMs at the T3 to T4 level, with the feeder vessel originating from the right thyrocervical trunk and flowing via the ASA. Following surgical treatment for a perimedullary AVF, she received fRT at a dose of 20 Gy in 4 fractions over consecutive days using the 3DCRT technique. At 20.9 months of follow-up, the MRI showed that the nidus had reduced in size, with no bleeding detected. However, an MRI scan at 57 months post-fRT showed a new aneurysm located at the T2 level. A review of the prior fRT plan confirmed that this aneurysm was located above the irradiated area. Although the cause of the new aneurysm cannot be fully determined, it is likely unrelated to the radiation therapy. Discussion As a result of the rarity of im-sAVMs, there is no established standard treatment regarding optimal modalities, their sequencing, or the definitive role of radiation therapy. Because of noninvasiveness, radiation therapy is theoretically an appropriate option in contrast with surgery or embolization, which carries a risk of severe, permanent treatment-related adverse events, such as spinal cord infarction. Additionally, details regarding radiation therapy, including the technique, dose, and fractionation, remain subjects of debate. Several articles have been published on the outcome of fRT as a part of im-sAVM treatment for more than a decade (Table 2).8, 9, 10, 11, 12, 13, 14, 15, 16 Most of these studies are small retrospective series or case reports. The largest study was reported by Kalani et al,13 which included 37 patients with im-sAVMs treated with fRT using the CyberKnife system. This study found a complete obliteration rate of 19% for the nidus, with no cases of rebleeding after fRT. However, the prescribed doses varied in this study, with a biologically effective dose (α/β = 2) (BED2) range of 70 to 120 Gy, and one patient developed radiation-induced myelopathy (1 of 37, 2.7%), making it difficult to adopt this protocol as a reference. Therefore, we conducted a retrospective review of our patients treated with our fRT protocol. Although there was a previously published result,9 the initial study included mostly conventional fractionation radiation therapy (70% used 1.8-2 Gy/d). With additional patients, a focus solely on fRT, continuous advancement in radiation therapy techniques, and extended follow-up, we now present the outcomes as one of the largest series on fRT in im-sAVMs.Table 2Summary of studies reporting details and outcomes of RT for treating im-sAVMs, including a comparison with our studyTable 2 dummy alt textStudiesNo.Median age (y)SexF: MHemorrhage before RTTypeTreatmentbefore RTDose(Gy/F) (BED2)Target volumeTechniqueF/UNidus (%)RMRebleedCRPRSDPDWolkov et al, 198882 (1963-1981)12, 752:0NRNR2 surgery45-50/25-28(85.5-125)7, 12.5 cmConventional5,12 yNR0%0%Hida et al, 2003910(1982-1999)25(15-50)6:4100%Intramedullary3 embolization2 Surgery32-40/20(57.6-80)32/16, 32.4/18(61.56,64)20/4,30/8(70,86.5)NR3DCRTMedian 49 mo(26-124)0505000%0%Sinclair et al, 20061015(1997-2005)NRNR53%Intramedullary7 embolization1 surgery25/5 (87.5)21/3-4(76.7-94.5)20/2 (120)Mean 2.36 ccCyberKnifestainless fiducialaverage isodose 83%Mean 27.9 mo(3-59)1/5*3/5*1/5*00%0%Gekka et al, 2014111(2007)370:1100%Juvenile typeNone20/4 (70)NRNR5 y1000%0%Potharaju et al, 2014123(2010-2011)26(25-57)2:1100%Intramedullary2 embolization21/3 (94.5)Mean 4.05 ccCyberKnifeVaclok, X-sight spine trackingIsodose80%-85%Mean35.7 mo(24-39)66.6033.300%0%Kalani et al, 20161337(1996-2014)30(9-56)19:1850%Intramedullary16 embolization7 surgery5 SBRT20/3-4(70-86.7)21/3 (94.5)20/2 (120)Mean 2.3 ccCyberKnifeuse fiducialMean 39.7 mo(2-121)19463503%0%Rashad et al, 2017145(2010-2014)39(31-70)1:440%Intramedullary(4 compact 1 juvenile)2 embolization18/3 (72)NRCyberKnifeMean 37.2 mo(16-62)0208000%0%Mori et al, 2018155(2006-2014)32(23-54)3:260%Intramedullary2 embolization20/4 (70)Mean 2.62 ccNovalisMedian5.5 y(3-8.5)40060020%(temporary)0%Suzuki et al, 20231610(2011-2022)17.5(7-55)3:790%Intramedullary1 embolization1 surgery14-24/2-7(37.5-94.5)Mean 0.72 ccCyberKnifeisodose 78.2%Mean 47.2 mo(6-125)0109000%0%Present study21(2003-2023)27(14-55)11:1042.8%Intramedullary3 embolization12 surgery20/4 (70)24/4 (96)1,7, 2.3 cc(IMRT patients)3DCRT 19IMRT 2Median56.4 mo(12-236)10384840%0%Abbreviations: NR = not reported; 3DCRT = 3-dimensional conformal radiation therapy; BED2 = biological effective dose (α/β = 2); CR = complete response; IMRT = intensity modulated radiation therapy; im-sAVMs = intramedullary spinal arteriovenous malformation; PD = progressive disease; PR = partial response; RM = radiation-induced myelopathy; RT = radiation therapy; SD = stable disease.⁎Only 5 patients were able to undergo imaging evaluation in this study. Table 2Summary of studies reporting details and outcomes of RT for treating im-sAVMs, including a comparison with our studyTable 2 dummy alt textStudiesNo.Median age (y)SexF: MHemorrhage before RTTypeTreatmentbefore RTDose(Gy/F) (BED2)Target volumeTechniqueF/UNidus (%)RMRebleedCRPRSDPDWolkov et al, 198882 (1963-1981)12, 752:0NRNR2 surgery45-50/25-28(85.5-125)7, 12.5 cmConventional5,12 yNR0%0%Hida et al, 2003910(1982-1999)25(15-50)6:4100%Intramedullary3 embolization2 Surgery32-40/20(57.6-80)32/16, 32.4/18(61.56,64)20/4,30/8(70,86.5)NR3DCRTMedian 49 mo(26-124)0505000%0%Sinclair et al, 20061015(1997-2005)NRNR53%Intramedullary7 embolization1 surgery25/5 (87.5)21/3-4(76.7-94.5)20/2 (120)Mean 2.36 ccCyberKnifestainless fiducialaverage isodose 83%Mean 27.9 mo(3-59)1/5*3/5*1/5*00%0%Gekka et al, 2014111(2007)370:1100%Juvenile typeNone20/4 (70)NRNR5 y1000%0%Potharaju et al, 2014123(2010-2011)26(25-57)2:1100%Intramedullary2 embolization21/3 (94.5)Mean 4.05 ccCyberKnifeVaclok, X-sight spine trackingIsodose80%-85%Mean35.7 mo(24-39)66.6033.300%0%Kalani et al, 20161337(1996-2014)30(9-56)19:1850%Intramedullary16 embolization7 surgery5 SBRT20/3-4(70-86.7)21/3 (94.5)20/2 (120)Mean 2.3 ccCyberKnifeuse fiducialMean 39.7 mo(2-121)19463503%0%Rashad et al, 2017145(2010-2014)39(31-70)1:440%Intramedullary(4 compact 1 juvenile)2 embolization18/3 (72)NRCyberKnifeMean 37.2 mo(16-62)0208000%0%Mori et al, 2018155(2006-2014)32(23-54)3:260%Intramedullary2 embolization20/4 (70)Mean 2.62 ccNovalisMedian5.5 y(3-8.5)40060020%(temporary)0%Suzuki et al, 20231610(2011-2022)17.5(7-55)3:790%Intramedullary1 embolization1 surgery14-24/2-7(37.5-94.5)Mean 0.72 ccCyberKnifeisodose 78.2%Mean 47.2 mo(6-125)0109000%0%Present study21(2003-2023)27(14-55)11:1042.8%Intramedullary3 embolization12 surgery20/4 (70)24/4 (96)1,7, 2.3 cc(IMRT patients)3DCRT 19IMRT 2Median56.4 mo(12-236)10384840%0%Abbreviations: NR = not reported; 3DCRT = 3-dimensional conformal radiation therapy; BED2 = biological effective dose (α/β = 2); CR = complete response; IMRT = intensity modulated radiation therapy; im-sAVMs = intramedullary spinal arteriovenous malformation; PD = progressive disease; PR = partial response; RM = radiation-induced myelopathy; RT = radiation therapy; SD = stable disease.⁎Only 5 patients were able to undergo imaging evaluation in this study. Summary of studies reporting details and outcomes of RT for treating im-sAVMs, including a comparison with our study Abbreviations: NR = not reported; 3DCRT = 3-dimensional conformal radiation therapy; BED2 = biological effective dose (α/β = 2); CR = complete response; IMRT = intensity modulated radiation therapy; im-sAVMs = intramedullary spinal arteriovenous malformation; PD = progressive disease; PR = partial response; RM = radiation-induced myelopathy; RT = radiation therapy; SD = stable disease. Only 5 patients were able to undergo imaging evaluation in this study. Our study demonstrated that fRT with a dose of 20 to 24 Gy in 4 fractions was both effective and safe, with no cases of rebleeding after treatment and no radiation-related myelopathy observed, even after more than 10 years of follow-up. Although the complete obliteration rate in our study appears lower compared to other studies, this may be because of the relatively low BED2 of 70 to 96 Gy. Nevertheless, the rebleeding rate was zero. In contrast, there is evidence that partial obliteration of the intracranial arteriovenous malformation (bAVMs) does not reduce the annual hemorrhage rate.17 As a result, the goal in treating bAVMs is complete nidus obliteration, which requires high-dose radiation therapy for definitive treatment. Flickinger et al18 analyzed the dose-response relationship between radiosurgery dose and nidus obliteration rates, finding that radiosurgery doses of 20 to 24 Gy achieved obliteration rates of 65% to 86%. However, such high doses cannot be used in cases of im-sAVMs because of spinal cord dose tolerance limitations.19, 20, 21 According to previous studies, including ours (Table 2), most of the series used radiation doses with a BED2 of less than 100 Gy, and no patients in these studies experienced rebleeding. This suggests that treatment of im-sAVMs may not require high doses to achieve complete obliteration, thereby reducing the risk of radiation-induced myelopathy. The difference in treatment response between im-sAVMs and bAVMs remains unclear because of the limited understanding of the causes and natural history of im-sAVMs. One possible explanation lies in the possible differences in blood flow and pressure between intracerebral and intramedullary vessels. Blood flow within the spinal cord may be lower than in the brain,22 allowing radiosurgery-induced flow reduction to be sufficient to prevent bleeding. In contrast, the higher blood pressure in intracerebral vessels may need complete obliteration of the nidus, as merely decreasing blood flow might not be enough. However, this theory lacks direct evidence. Furthermore, most studies on spinal cord blood flow have been conducted in animal models because of the technical challenges of measuring or sampling venous drainage from the small spinal cord venous system.23 Future advancements in technology for measuring human spinal cord blood flow, combined with accumulated treatment outcomes from more patients, may help clarify this issue. Another important aspect to discuss is the fRT technique. Although many institutions use radiosurgery systems equipped with image guided radiation therapy modalities to correct intrafraction movement, our experience treating patients with im-sAVMs suggests that immobilization with a shell or cushion, combined with bony landmarks for image guidance, is sufficient. This approach has enabled us to deliver effective and safe treatment without the need for costly equipment. This method may be beneficial for nontertiary care centers where specialized treatment systems are not available. Limitation Our study has a few limitations. Because radiation therapy techniques and treatment systems have evolved over time, only 2 patients in this study were treated with the IMRT technique. IMRT provides a highly conformal dose distribution, making it well-suited for fRT planning. Additionally, as im-sAVMs typically occur in younger patients and the condition is benign, concerns about secondary malignancy or tumor development should be more than usual. IMRT has the advantage of reducing radiation exposure to normal tissues by decreasing the irradiated target volume. Several studies have reported that using radiosurgery with advanced treatment systems does not increase the incidence of secondary malignancy or malignant transformation in benign central nervous system tumors compared to the general population and shows a lower incidence than conventional or 3-field radiation therapy.24, 25, 26 However, with only 2 patients treated with advanced IMRT in our study, we were unable to report certain data, such as GTV nidus volume, or to evaluate the clinical advantages of IMRT over older 3DCRT planning. Another limitation is that each patient received other treatments before and/or after radiation therapy (57.1% and 14.3% of the patients received microsurgery and embolization before fRT, respectively, whereas 4.8% underwent microsurgery for the residual coexisting AVF after fRT), preventing us from isolating the effect of radiation therapy alone on rebleeding prevention or accurately assessing its impact on the annual bleeding rate compared to others. Although a prospective or randomized controlled study could address these limitations, conducting such a study is challenging because of the rarity of this disease. As of now, all available literature, including our study, consists of retrospective reviews and represents the current best evidence. Conclusions Fractionated radiation therapy at doses of 20 to 24 Gy in 4 fractions for im-sAVMs has been shown to be safe and may effectively reduce the risk of bleeding, with no severe permanent radiation-induced myelopathy observed even after long-term follow-up. We conclude that fRT is a reasonable option, either as an adjunct to other modalities or as a primary treatment choice when other treatments are not feasible. Disclosures The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.