EANO Consensus Statement on Radiation Necrosis

Consensus

EANO Consensus Statement on Radiation Necrosis

ForPost-cranial-RT pts with a new enhancing lesion, glioma or brain metastases

TL;DRDelphi consensus: 53/57 statements reached ≥80% agreement across 20 experts; perfusion MRI + amino acid PET preferred, bevacizumab for steroid-refractory RN.

Why it mattersRadiation oncology

The dose-volume numbers are the transferable part: V12Gy ~10 cm³ carrying 5-10% symptomatic necrosis in single-fraction SRS, and hypofractionated brain+target V20Gy (3fx) or V24Gy (5fx) <20 cm³ holding RN needing resection under 4%. That is a planning constraint, and it gates the single-session vs hypofractionated choice for larger or eloquent-region metastases.

Monday clinic

For a patient 6-24 months out from SRS or cranial RT with a new enhancing lesion, this supports going to perfusion MRI plus amino acid PET before calling progression; it does not settle mixed lesions, where the panel still points to tissue.

11 details 5 trials watching

Design

Three-round Delphi. 57 statements drafted after literature review across epidemiology-pathophysiology (n=8), causes (n=10), imaging (n=10), histopathology (n=9), management (n=20), plus an exploratory research set of 8. Consensus prespecified at ≥80% agreement on a 5-point Likert scale.

Population & inclusion

The voting panel, not patients: 15 volunteers (2 radiation/clinical oncologists, 4 neuro-oncologists, 4 medical oncologists, 2 neurosurgeons, 1 neuropathologist, 1 neuroradiologist, 1 nuclear medicine physician), expanded by 5 more in round 1 (4 neuro-oncologists, 1 radiation/clinical oncologist) to 20 total. Clinically the scope is RN after RT for primary glial or metastatic brain tumors.

Radiotherapy

RN is called uncommon below EQD2 ~40 Gy (α/β 2 Gy), rising with dose per fraction, total dose and irradiated volume. V12Gy ~10 cm³ carries a 5-10% symptomatic necrosis risk after single-session SRS; for hypofractionated SRT, brain-plus-target V20Gy (3fx) or V24Gy (5fx) <20 cm³ associates with <10% any necrosis and <4% RN requiring resection. Re-irradiation risk stays under 10% at cumulative EQD2 100-110 Gy and reaches up to 25% above 130 Gy.

Results

Consensus reached on 53 of 57 statements after three rounds. Diagnostic and management statements are tabulated separately in the details; the four that failed to reach 80% are not identified in the available text.

Modality	Setting	Sensitivity	Specificity
T1 contrast-enhanced MRI alone	WHO grade 3-4 glioma	68%	77%
T1 contrast-enhanced MRI alone	Brain metastases	79%	76%
DSC rCBV	Grade 3-4 glioma	87-90%	86-88%
DSC rCBV	Brain metastases (418 lesions)	83%	78%
MR spectroscopy Cho/NAA	Glioma (455 pts)	88%	86%
Amino acid PET	Brain metastases (~400 pts)	82%	84%

Discussion

The document's practical contribution is a decision sequence rather than a new treatment: advanced MRI first, amino acid PET to add confidence, tissue when doubt persists, and treatment triggered only by symptoms of >7 days with increasing severity. Its most consequential position is moving bevacizumab from rescue to an expected next step at a defined steroid threshold (dependence >4 weeks or dexamethasone 8 mg/day), which is a lower bar than many centres apply.

Applies to

symptomatic cerebral RN after RT for glioma or brain metastases, in centres with perfusion MRI and amino acid PET access

Does not represent spinal or head-and-neck RN, pseudoprogression within the first months after chemoradiation, or the late enlarging cysts after SRT that the panel says are pathogenetically separate.

Limitations

Agreement percentages measure how a self-selected 20-person European panel voted, and the panel skews neuro-oncology (8 of 20) over radiation oncology (3 of 20) for a complication defined by dose and volume. FET and F-DOPA availability is largely European, so the imaging algorithm's top tier is not reproducible in most centres worldwide. The 4 non-consensus statements are not enumerated in the accessible text, so the reader cannot see where the panel actually split.

Rationale

Delphi expert-opinion document, no efficacy endpoint. Agreement measured against opinion, not outcomes; authors state absence of Level 1 evidence and call for randomised trials.

Open questions

Optimal bevacizumab dose, interval and duration for symptomatic RN

recruiting Bevacizumab Versus Corticosteroids as First-line Treatment in Patients With Symptomatic Cerebral Radiation Necrosis After Radiation for High-grade Glioma or Brain Metastases Phase 3

n=408 · primary completion 2028-07 · phase 3 bevacizumab vs dexamethasone, 1L sCRN

recruiting Corticodependent or Corticoresistant Brain Radionecrosis After Radiotherapy for Brain Metastases Phase 3

n=84 · primary completion 2028-08 · randomised bev vs placebo in steroid-refractory RN
Single-session SRS vs hypofractionated SRT for RN risk

active Comparing Single vs Multiple Dose Radiation for Cancer Patients With Brain Metastasis and Receiving Immunotherapy Phase NA

n=58 · primary completion 2028-03 · single vs fractionated SRS toxicity on ICI

recruiting Brain Metastases in Greater Size - Hypofractionated Options Trial (BIGSHOT) Phase 2

n=54 · primary completion 2029-02 · staged SRS vs FSRT, 2-5 cm mets, safety endpoint

recruiting Stereotactic Radiotherapy in Oligometastatic Brain Disease: a Randomised Phase III Study Comparing Hypofractionated Stereotactic Radiation Therapy (3*10 Gy) to the Historical Single-dose Radiosurgery (1*20 to 25 Gy) With Medico-economic Evaluation. Phase NA

n=504 · primary completion 2030-01 · phase 3 3x10 Gy SRT vs 1x20-25 Gy SRS
Predictors of bevacizumab response and recurrence rate after treatment

📚 Sources · 📄 1 paper

📄 PAPER Duerinck, Johnny; Van Den Bent, Martin; Brandal, Petter et al. · Neuro-Oncology (2026-07)

The European Association for Neuro-oncology (EANO) Consensus Statement on Radiation Necrosis

doi

Abstract

Abstract Introduction Radiation necrosis (RN) complicates neuro-oncological care, mimicking tumor recurrence and lacking high-level evidence for standardized management. Methods A European Association for Neuro-Oncology (EANO) expert panel utilized a three-round Delphi process to create a comprehensive expert opinion document based on the available current scientific evidence. A series of statements, derived from the published literature were created by the experts in each field. Consensus was defined as ≥ 80% agreement using a 5-point Likert scale. Results After three rounds among 20 experts that included adaptation of statements, the Delphi process reached a consensus (≥80% agreement) on 53 statements out of 57. RN occurs in 4% to 30% of patients, typically appearing 6 to 24 months after radiotherapy for primary (glial) or metastatic brain tumors. Experts identified perfusion MRI and amino acid PET as the most suitable imaging modalities for differentiation from tumor recurrence. While histopathology remains the gold standard, identifying viable tumor cells in irradiated gliomas is challenging due to overlapping cytological features with reactive glia. For symptomatic management, corticosteroids may be tried, and bevacizumab is recommended for corticosteroid-refractory cases, with evidence suggesting profound efficacy even at low doses. Surgery is considered effective for rapid symptom relief and definitive diagnosis in accessible lesions. Laser Interstitial Thermal Therapy (LITT) can be considered an additional treatment option for symptomatic RN. Conclusions Despite the absence of Level 1 evidence, these Delphi-survey-formulated recommendations provide actionable guidance for clinical practice. There is a need for prospective randomized trials focusing on symptomatic RN.

Radiation Supportive Consensus / guideline Consensus

The longer read

The value of this document is not that it settles anything about radiation necrosis, because it cannot: the panel says plainly there is no Level 1 evidence and closes by asking for prospective randomised trials in symptomatic RN. What it does is fix a defensible default at each of the three decision points where practice currently diverges most, and it is worth reading for where those defaults sit relative to what many centres actually do.

The first is diagnostic. Placing perfusion MRI and amino acid PET in sequence rather than in competition is well supported by the numbers the panel assembles: contrast-enhanced T1 alone reaches only 68% sensitivity and 77% specificity in grade 3-4 glioma, while DSC rCBV reaches 87-90% and 86-88% in the same setting. The honest caveat is that these are pooled figures from retrospective and imaging-focused series, where the reference standard is often follow-up rather than tissue, and where lesions too ambiguous to classify tend not to enter the denominator. A statement that combined advanced MRI plus PET has the highest accuracy carries 94.4% panel agreement but rests on very little head-to-head data. The recommendation is still the right default; a reader should just not expect the 90% figures to hold on the ambiguous mixed lesion sitting in front of them, which is precisely the case where imaging is being asked to do the most work.

The second is the steroid-to-bevacizumab threshold, and this is where the document is most likely to change behaviour. Defining escalation at steroid dependence beyond 4 weeks, or a requirement for dexamethasone 8 mg daily, converts a judgment call into a trigger. That threshold earned 87.5% agreement, on the lower end of this panel's range, which is a fair reflection of the evidence underneath it: the panel simultaneously concedes that duration, interval and optimal dose are unknown, that no dosing scheme has evidence of superiority (four are listed, from 5 mg/kg q2w to fixed 600 mg q3w), that predictors of response are undetermined, and that the recurrence rate after treatment is unknown. So the trigger is specified and everything downstream of it is not. For a radiation oncologist the practical read is that reaching for bevacizumab earlier is now defensible, while what happens after four infusions remains entirely local practice.

The third is the dosimetric section, which is the part that transfers directly into planning. Framing RN as uncommon below EQD2 40 Gy, naming V12Gy near 10 cm³ as the single-fraction inflection at 5-10% symptomatic risk, and giving hypofractionated constraints of V20Gy in 3 fractions or V24Gy in 5 fractions under 20 cm³, restates the pooled 51-study analysis rather than adding to it. Its usefulness is that it is now written into a consensus document that a tumor board will accept. The re-irradiation figures matter more, because the jump from under 10% at cumulative EQD2 100-110 Gy to as much as 25% above 130 Gy is steep enough to be a genuine constraint on repeat SRS, and the panel pairs it with the observation that mean diameter above 2.0-3.0 cm carries higher risk. The four ongoing randomised trials of single-session SRS versus hypofractionated SRT that the paper cites are what will actually answer this; until they read out, the constraints here are the best available and should be treated as descriptive of risk, not as validated limits.