Adverse Radiation Effects

1.The dominant late toxicity of cranial radiosurgery

Radiation necrosis is the most consequential late effect of cranial SRS. It typically declares itself 3–18 months after treatment, with a median around 7 months and a tail extending years later, as a focally enhancing, edematous lesion at the treated site. The strongest dosimetric predictor is the volume of normal brain receiving ≥12 Gy (V12Gy). For single-fraction SRS to brain metastases, a V12Gy of 5, 10, and more than 15 cc corresponds to symptomatic radionecrosis risks of roughly 10%, 15%, and 20%, and a common single-fraction planning objective is V12Gy at or under 10 cc; for 3-fraction regimens the corresponding proposals are V18Gy under 30 cc and V23Gy under 7 cc. Risk climbs steeply as V12 rises, which is why staged or fractionated SRS is used for larger targets. Prior whole-brain radiotherapy, re-irradiation, and larger target volume all raise the risk, and prior SRS to the same site carries the steepest penalty, with roughly a 20% one-year risk of symptomatic effect versus 3% with no prior treatment. Among systemic agents, dual immune-checkpoint blockade given close to SRS is the strongest current signal, with symptomatic necrosis in about 26% of patients versus roughly 12–14% with single-agent or no checkpoint inhibitor; separating radiosurgery and dual checkpoint blockade by more than 4 weeks brings the rate back into line. VEGFR and EGFR tyrosine kinase inhibitors, vemurafenib, and antibody-drug conjugates have also been associated with higher necrosis rates, and consensus guidance recommends holding BRAF and MEK inhibitors for at least 1 day before and after radiosurgery. Reported symptomatic rates are on the order of 5–15%, higher for large lesions and re-treatment.

2.The central diagnostic problem

A new or enlarging enhancing lesion after SRS for a metastasis or glioma may be necrosis or recurrent tumor, and the two look alike on conventional MRI. Distinguishing them changes management entirely, so it leans on advanced imaging:

• MR perfusion (DSC, relative cerebral blood volume): rCBV is higher in recurrent tumor and lower in necrosis; a meta-analysis of 13 studies confirmed significantly higher rCBV in recurrence than in radiation injury but did not establish a cutoff, and reported thresholds vary by pathology and perfusion technique. Representative single-institution values include an rCBV ratio above 2.1 favoring recurrent brain metastasis and a cutoff of 1.75 that performed only moderately in recurrent glioblastoma.
• MR spectroscopy: elevated choline and Cho/NAA and Cho/Cr ratios favor tumor; necrosis shows a depressed, disorganized spectrum.
• Amino-acid PET (FET, FDOPA, MET) is recommended alongside MRI to separate recurrent brain metastasis from treatment-related change, with pooled sensitivity and specificity around 82% and 84%; typical tumor-to-background thresholds are 1.6 for FET and 1.3 for MET. Increased uptake favors viable tumor, and necrosis typically shows low or absent uptake. Somatostatin-receptor PET with DOTATATE has a role in meningioma but is not part of the brain-metastasis necrosis workup.
• Lesion quotient (T1/T2 matching): the ratio of the nodule area on T2 to the enhancing area on post-gadolinium T1 on a comparable axial slice. A value of 0.3 or less was originally described as indicating necrosis, but attempts to validate the sign have shown poor positive predictive value, so it should not drive management on its own.
• Serial MRI with a growth-then-plateau-then-regression pattern favors necrosis; relentless progression favors tumor.
• Biopsy / resection remains the arbiter when imaging is equivocal and the answer drives therapy, and frequently the specimen shows a mix of both.

3.A stepwise ladder, with evidence tiers

Treatment is matched to symptoms and mass effect. Asymptomatic imaging change is observed; symptomatic edema or necrosis is treated, escalating from steroids through anti-VEGF therapy to procedures.

• Observation, for asymptomatic imaging change; most early post-SRS enhancement/edema is followed, not treated.
• Corticosteroids (dexamethasone): first line for symptomatic edema. Dose to symptom severity and mass effect, then taper as soon as clinically feasible; monitor glucose, infection, myopathy, mood, and gastrointestinal risk. There is no single regimen appropriate for every patient.
• Bevacizumab: the best-supported steroid-sparing drug for symptomatic, steroid-refractory necrosis, backed by two randomized trials: a double-blind crossover trial of 14 patients in which all bevacizumab-treated patients improved radiographically and clinically and no placebo patient did, and an open-label trial of 112 patients with radiation-induced brain necrosis after nasopharyngeal carcinoma radiotherapy, in which 5 mg/kg every 2 weeks for 4 cycles produced a 65.5% two-month response rate versus 31.5% with corticosteroids. Reported regimens range from 5 to 10 mg/kg every 2 to 3 weeks for 2 to 6 cycles, with no schedule established as superior. Consider bevacizumab when a patient requires 8 mg or more of dexamethasone daily for longer than 4 weeks; balance radiographic and clinical benefit against hypertension, proteinuria, thrombosis, bleeding, and impaired wound healing, and recognize that recurrence can occur after discontinuation.
• Boswellia serrata (boswellic acids): an optional, investigational oral anti-edema adjunct, not a rung equivalent to bevacizumab. Reported dosing is roughly 4200–4800 mg per day, but evidence in established post-SRS necrosis remains limited and product formulation and bioavailability vary.
• Pentoxifylline + vitamin E and hyperbaric oxygen: occasionally used adjuncts with low-quality CNS evidence. They should not delay bevacizumab, tissue diagnosis, LITT, or resection when symptoms or mass effect require definitive treatment.
• LITT (laser interstitial thermal therapy): minimally invasive cytoreduction for focal, surgically accessible necrosis, and a route to tissue diagnosis.
• Surgical resection, for refractory symptoms, significant mass effect, or diagnostic uncertainty; definitively relieves the lesion and yields tissue.

4.Adverse radiation effects after AVM radiosurgery

After AVM SRS, post-radiosurgery imaging changes (perinidal T2/FLAIR signal and enhancement) are common, radiologically in roughly a third of patients, but symptomatic in a smaller minority (on the order of 10%), usually transient and steroid-responsive. Late cyst formation and chronic encapsulated hematoma are recognized delayed effects years out. These are distinct from the latency-period hemorrhage risk, the AVM can still bleed until obliteration is achieved, which is a feature of the unobliterated nidus, not a radiation injury.

5.Radiation-induced optic neuropathy, hypopituitarism, myelopathy

• Radiation-induced optic neuropathy (RION): painless, often sudden visual loss, typically months to ~3 years out and frequently irreversible. It is constraint-driven and largely preventable: keep the optic apparatus maximum point dose below ~8 Gy in a single fraction, where risk is negligible; risk remains low (on the order of ~1%) up to ~10–12 Gy and rises steeply above that. There is no reliable treatment once it occurs (corticosteroids, anticoagulation, bevacizumab, and hyperbaric oxygen have all been tried with limited success) so prevention by constraint is paramount, and prior radiation lowers tolerance.
• Hypopituitarism after sellar/parasellar SRS develops gradually, accruing to a long-term incidence commonly cited around 20–40% by 10 years (higher with greater stalk and gland dose). The growth-hormone and gonadal axes tend to fail earliest, followed by ACTH and TSH; lifelong annual endocrine screening is required, and keeping dose to the pituitary stalk and gland as low as feasible reduces risk.
• Radiation myelopathy after spine SBRT is rare (well under 1% when cord constraints are respected) but catastrophic and usually irreversible, presenting as a delayed (months–years) myelopathy. Cumulative cord dose governs the reirradiation setting, where limits tighten and prior dose must be accounted for. As with RION, there is no reliable treatment, so the constraint is non-negotiable.

6.Vertebral compression fracture, pain flare, cranial neuropathy

• Vertebral compression fracture (VCF): the signature structural toxicity of spine SBRT, with a pooled crude incidence near 14% and individual series reporting 11–39%, reaching about 39% after 24 Gy in a single fraction. It occurs early rather than late: median time to fracture is about 2.5 months and roughly two thirds occur within 4 months, so imaging surveillance should start early. Risk rises with a dose per fraction of 20 Gy or more, baseline vertebral collapse or fracture, lytic tumor, and spinal malalignment, which are individual SINS criteria rather than the composite score. Painful fractures are managed with cement augmentation (vertebroplasty or kyphoplasty); instability or deformity needs surgical stabilization. A high SINS up front should prompt consideration of prophylactic stabilization before, or instead of, ablative single-fraction dosing.
• Pain flare: a transient post-treatment pain increase affecting a majority of unprotected patients after spine SBRT, reported in about 68% of steroid-naive patients, usually within days. Prophylactic dexamethasone cuts this to roughly 19%: a common regimen is 4 mg given 1 hour before the first fraction and continued for 4 days (some use 8 mg); breakthrough flare is treated with a short steroid course.
• Cranial neuropathy: after vestibular schwannoma SRS, transient trigeminal or facial dysfunction and dose-dependent hearing loss: the cochlear mean dose matters, with a goal of roughly ≤ 4 Gy for hearing preservation; after cavernous-sinus targets, cranial-nerve tolerance is generally favorable. Brainstem toxicity is constraint-limited (Dmax ≈ 12.5–15 Gy single fraction).