The dilemma of radiation necrosis from diagnosis to treatment in the management of brain metastases

Abstract

Radiation therapy with stereotactic radiosurgery (SRS) or whole brain radiation therapy is a mainstay of treatment for patients with brain metastases. The use of SRS in the management of brain metastases is becoming increasingly common and provides excellent local control. Cerebral radiation necrosis (RN) is a late complication of radiation treatment that can be seen months to years following treatment and is often indistinguishable from tumor progression on conventional imaging. In this review article, we explore risk factors associated with the development of radiation necrosis, advanced imaging modalities used to aid in diagnosis, and potential treatment strategies to manage side effects.

Brain metastases are the most common intracranial neoplasm in adults with cancer, and it is estimated that up to 30–40% of patients with solid malignancies will develop intracranial involvement during the course of their disease.¹ With continued improvements in systemic therapies, cancer patients are surviving longer and as a result the incidence of brain metastases will continue to increase. Stereotactic radiosurgery (SRS) is a standard of care treatment for brain metastases, with the American Society for Radiation Oncology (ASTRO) recommending strong support for SRS for patients with ≤4 brain metastases, with conditional recommendations for SRS in those with 5–10 lesions.¹ Though SRS provides excellent local control for brain metastases, cerebral radiation necrosis (RN) is a late complication that may be seen months to years after treatment. The development of RN is a multifactorial process that occurs when radiation results in vascular injury and damage to glial and endothelial cells, ultimately resulting in tissue necrosis. Radiation necrosis is oftentimes indistinguishable from tumor progression (TP), with a variable clinical presentation, ranging from asymptomatic and identified on routine surveillance to progressive symptoms resulting in significant morbidity. Accurately distinguishing RN from TP using imaging modalities such as magnetic resonance imaging (MRI), magnetic resonance spectroscopy (MRS), or positron emission tomography (PET) is important as the appropriate diagnosis is needed to guide management recommendations.² Several treatment options exist, and typically corticosteroids are the first-line therapy. In this article, we review the incidence, presentation, risk factors, diagnostic modalities, and treatment options for RN following radiation therapy (RT) for brain metastases.

Incidence

The exact incidence of radiation necrosis following radiation therapy to the brain is unknown, with reported rates ranging from 0% to 30%.³ The reasoning for this wide range is multifactorial, potentially reflecting the differences in diagnostic definitions and radiographic techniques used throughout various studies. The gold standard for accurate diagnosis is pathologic assessment, but given the potential complications associated with biopsy or surgery, this is infrequently performed, and diagnosis is made based on radiographic findings. In the seminal Radiation Therapy Oncology Group (RTOG) 9005 SRS dose escalation study of recurrent brain metastases or gliomas, the 1- and 2-year incidence of radiation necrosis was 8% and 11%, respectively.⁴ In the early Italian experience investigating factors associated with toxicity from SRS, they reviewed 310 treated lesions and found a RN rate of 24%, with 10% of lesions being symptomatic while 14% were asymptomatic.⁵ Similarly, the MSKCC experience of 271 metastases treated with single fraction SRS reported a rate of 26% at a median follow-up of 17 months, with 17% being symptomatic.⁶ However, a more modern prospective randomized trial of 213 adults with 1–3 brain metastases treated with SRS alone versus SRS with whole brain radiation therapy (WBRT) found an incidence of only 4.5% and 2.9%, respectively.⁷ The heterogeneity in reported rates highlights the need for advances in diagnostic techniques.

Presentation and Risk Factors

Following treatment of brain metastases with SRS, radiation-induced injury can be categorized as acute, early delayed (pseudoprogression), or late. RN is considered a late complication of RT and is typically seen months to years following treatment. Symptoms of RN mimic that of tumor recurrence and are dependent on intracranial location, and may include motor deficits, sensory deficits, or changes in vision or speech; generalized symptoms of headache, nausea, and somnolence are most commonly seen.⁵

Histology

Specific histological subtypes appear to be at increased risk of RN. For example, in our institutional experience of 5747 brain metastases, 427 lesions developed RN. Histological subgroup analysis showed that lung adenocarcinoma histology, renal histology, ALK rearrangement, HER2 amplification, and BRAF V600 + mutation were associated with RN.⁸

Location

Few studies have identified intracranial locations that pose an increased risk for developing RN. However, Korytko et al. found that occipital and temporal lesions are at increased risk.⁹ A study of AVM treated with SRS found that lesions of the pons/midbrain were at the highest risk, while frontal lobe and temporal lesions had the lowest risk.¹⁰

Radiation Dose and Volume

RTOG 9005 investigated the maximum tolerated dose for recurrent gliomas and brain metastases, and RN was seen in 8% of patients at 1 year and 11% at 2 years. Conformality index (prescription isodose volume/tumor volume) >2 and homogeneity index (maximum dose/prescribed dose) >2 contributed to grades 3–5 toxicities. Larger tumors had a higher risk of RN, with the report recommending doses for SRS of 24 Gy for tumors ≤2 cm, 18 Gy for 2.1–3.0 cm, and 15 Gy for 3.1–4.0 cm.⁴ Sneed et al. also found larger tumors to be at increased RN risk, with tumors <1.5 cm having a ≤3% risk while tumors ≥1.6 cm had a ≥10% risk.¹¹

The volume of normal brain parenchyma receiving 10 Gy (V10 Gy) and 12 Gy (V12 Gy) with single fraction SRS has been shown to increase RN risk in multiple studies.^5,9,12,13 The initial experience by Korytko et al. for non-AVM tumors receiving SRS included 198 tumors and found the rate of symptomatic RN increased as the V12 increased; tumors >10 cc had a >50% risk of developing RN.⁹ Minniti et al. published their experience of 206 patients with 310 cerebral metastases undergoing single fraction SRS and reported RN risk of 47% in patients with V12 Gy>10.9 cm³, but only 10% if V12 Gy was <8.5 cm³.⁵ Similarly, Blonigen et al. suggested that V12 Gy <8.5 cm³ reduces RN risk.¹² In a recent multi-institutional report of SRS in patients with non-small cell lung cancer, melanoma, or renal cell carcinoma treated with single fraction SRS and immune checkpoint inhibitors, the authors suggested a V12 Gy <12 cm³ had a RN risk of 6.6%, whereas V12 Gy >20 cm³ had a risk of 20.3%.¹³ The high dose per fraction, hypofractionated treatment effects in clinic (HyTEC) report for single fraction SRS found a V12 Gy to 5 cm³, 10 cm³, or >15 cm³ was associated with an approximate 10%, 15%, and 20% risk of RN.¹⁴ Radiation hotspots > 110% within the planning target volume have also been found to increase risk.¹⁵

A potential strategy to mitigate the risk of RN in larger lesions is fractionated SRS, typically 25–40 Gy delivered over 3–5 fractions.¹⁴ The HyTEC report found V20 Gy <20 cm³ and V24 Gy <20 cm³ were associated with a <10% risk of RN or edema for 3 fractions and 5 fractions SRS, respectively.¹⁴

Re-irradiation

A history of prior WBRT or SRS increases RN risk.⁹ Sneed et al. investigated 435 patients with 2200 brain metastases and found the most important risk factor for RN to be prior SRS to the same lesion, with a 20% 1-year risk; in comparison, the risk of RN was 4% in the setting of prior WBRT and 8% with concurrent WBRT.¹¹ The risk of RN was only 3% if there was no prior RT. In a more recent study investigating adverse radiation effects in 229 brain metastases treated with repeat single fraction SRS, Sneed et al. found a 30% risk of radiation effect on brain MRI; however, the risk of symptomatic radiation effect was only 11%.¹⁶

Hypofractionated SRS, typically in 3–5 fractions, may be beneficial in the re-irradiation setting to potentially reduce the risk of RN. Still, in a series of 91 patients with 120 metastases, Yan et al. found that salvage hypofractionated SRS resulted in a RN risk of 15.6% and a symptomatic RN risk of 7.0% at 12 months.¹⁷ These studies highlight the need for new therapeutic strategies in the salvage setting given the increased risk of RN with re-irradiation.

Chemotherapy

In the Sneed et al. study discussed above, the receipt of capecitabine within 4 weeks of SRS increased the risk of RN.¹¹ In our institutional experience investigating the risk of RN with concurrent cytotoxic chemotherapy, chemotherapy was not found to increase the risk.¹⁸ Shen et al. investigated 291 SRS treatments in 193 patients, of which 108 treatments were also treated with concurrent systemic therapy; 46% of these were treated with myelosuppressive chemotherapy.¹⁹ RN risk was not increased with the use of myelosuppressive chemotherapy.

Targeted Therapy

The use of several targeted therapies appears to increase the risk of RN. In our institutional experience discussed above, Kim et al. showed VEGF receptor tyrosine kinase inhibitors (TKIs) increased the risk of RN (14.4% vs 6.6%) as did EGFR TKIs (15.6% vs 6.0%).¹⁸ The use of HER2 antibody therapy also increased risk (9.0% vs 5.3%). Similarly, Juloori et al. showed that TKIs within 30 days of SRS increased the risk of RN in patients with renal cell carcinoma (10.9% vs 6.4%).²⁰ Treatment with SRS following resistance to TKI drugs may also play a role in increased RN risk.²¹ In a small study of 45 patients, trastuzumab emtansine (T-DM1) increased the risk of RN by 13.5-fold.²² Likewise, Id Said et al. found that T-DM1 increases the risk of RN following SRS, as 30% of patients treated with T-DM1 in their cohort developed RN.²³

Immunotherapy

The significance of concurrent immunotherapy and RN has mixed results.^24–28 For example, in a study of 260 patients with 623 brain metastases, the use of concurrent immune checkpoint inhibitors did not result in increased RN.²⁹ Our institutional experience of SRS treated with concurrent anti-CTLA-4 or anti PD-1 also did not show an increased risk of RN.¹⁸ Conversely, Martin et al. found that receipt of immunotherapy increased risk of symptomatic RN, though this was not studied based on timing from immunotherapy to SRS.²⁴

Preoperative vs Postoperative SRS

It is suggested that preoperative SRS reduces the risk of RN compared to postoperative SRS because a smaller amount of normal brain tissue is radiated and the majority of high dose volume is resected at the time of surgery.³⁰ NRG BN012 (NCT05438212) is a phase III trial evaluating this question.

Intact Brain Metastasis vs Surgical Cavity SRS

There is growing evidence that the risk of radiation necrosis is reduced when SRS is delivered postoperatively versus to an intact brain metastasis. For example, in a study of 187 consecutive patients treated with hypofractionated SRS in 5-fractions, Faruqi et al. found that the risk of adverse radiation events was increased when treating an intact brain metastasis compared with the surgical cavity.³¹ Interestingly, the surgical cavity group had a 3.7 fold larger PTV, suggesting additional factors in addition to volume of normal tissue irradiated play a role in the development of RN.

Diagnosis

Patients with intracranial metastatic disease treated with radiation therapy require close follow-up surveillance imaging to assess the response of the treated lesions as well as to rule out development of new lesions. The National Comprehensive Cancer Network (NCCN) guidelines recommend an MRI of the brain every 2–3 months for the first 1–2 years, followed by an MRI of the brain every 4–6 months indefinitely.³² Following SRS, up to one-third of treated lesions will increase in size.³³ Thus, appropriate diagnosis of RN versus TP is critical as this will guide next steps in management and avoid unnecessary treatments, such as surgery, re-irradiation, or discontinuation of a systemic therapy that was in fact effective.

Pathologic Assessment

Histopathology is the gold standard for accurate diagnosis, but may not always be feasible due to potential complications from surgery. Pathologic assessment may be accomplished by resection, biopsy, or during minimally invasive techniques such as laser-interstitial thermal therapy (LITT). LITT is discussed in more detail in the treatment section. Pathologically, RN is related to damage to endothelial and glial cells, resulting in vascular hyalinization and thrombosis with associated fibrinoid necrosis and hemorrhage.^34–36 The necrosis is typically paucicellular and surrounded by gliotic brain tissue; within the lesion, foamy macrophages and hemosiderophages may also be seen with dystrophic calcification. Contrary to RN, recurrent tumor possesses high cellularity with a high nuclei:cytoplasm ratio. Despite the differences in their appearance, histology does not always provide a clear diagnosis due to the presence of a mixed pattern of RN and viable tumor in the tissue sample.

Magnetic Resonance Imaging

Conventional MRI is a widely available tool with multiple imaging sequences, most notably T1-weighted with contrast and T2/FLAIR (fluid-attenuated inversion recovery). This results in excellent anatomic and spatial detail. Radiation necrosis typically appears as a ring-enhancing lesion on T1-weighted imaging and reflects a disruption in the blood-brain barrier, with surrounding T2/FLAIR signal, which represents vasogenic edema. Unfortunately, this is non-specific and can also be seen in the setting of tumor recurrence or infection.

Due to uncertainties in accurate diagnosis, several methods on conventional MRI sequences have been proposed to aid in the diagnosis of RN. A commonly available sequence, diffusion-weighted imaging (DWI) with apparent diffusion coefficient (ADC), measures the random motion of water molecules in tissue. Highly cellular tumor possesses a low ADC while ADC is elevated in RN. Early studies suggested that the lesion quotient (LQ), which is a ratio of the tumor nodule on T2-weighted imaging to the enhancing area on T1-weighted imaging, could accurately differentiate RN from TP.³⁷ However, this technique was not reproducible on subsequent studies with a sensitivity of 8% and specificity of 91% using the same definition.³⁸ Another method on conventional MRI is T1/T2 matching, which compares the volume of T1-weighted contrast enhancement with the borders on T2-weighted imaging. In a study of 68 patients who underwent resection following SRS, the authors concluded that well-correlated margins were consistent with TP while poorly correlated margins were more likely RN, with a T1/T2 mismatch providing a sensitivity of 83% and specificity of 91%.³⁹ Time-dependent changes in lesion morphology on conventional MRI following contrast administration may also help differentiate RN from TP, with RN demonstrating a nonenhancing interior on subtraction imaging of the 15 minute minus 55 minute time points and TP demonstrating enhancement.⁴⁰ Growth dynamics of lesion volume on consecutive conventional T1-weighted imaging may also be helpful. This was demonstrated in a study of 101 enlarging brain metastases, where RN was found to have larger growth dynamics compared to TP.⁴¹ However, the clinical utility of these methods is rarely used in practice, and at our institution we do not use these methods.

Radiomics

Radiomic signatures are an evolving field of artificial intelligence that builds predictive models by extracting large amounts of radiographic features. Radiomic signatures utilizing conventional T1-weighted, T2-weighted, and FLAIR sequences have been investigated and demonstrate promise. For example, in an analysis of 24 pathologically confirmed cases of RN and 73 cases of TP, the use of 5 radiomic features resulted in an accuracy of 73%.⁴²

Magnetic Resonance Perfusion

If conventional imaging does not provide a clear diagnosis, our institution next favors the use of relative cerebral blood volume (rCBV), a technique providing information on blood volume, blood flow, and permeability (Figure 1). It is postulated that since a viable tumor has intact vasculature, there is increased neovascularization with higher perfusion and blood volume, resulting in an increased rCBV in the setting of tumor but not in the setting of RN. For example, in one study evaluating 27 patients with brain metastases undergoing SRS, patients with tumor recurrence had rCBV ranging from 2.1 to 10 while those with RN were 0.39–2.57. The authors concluded that an rCBV cutoff of 2.1 was optimal, providing a specificity of 95% and a sensitivity of 100%.⁴³ In another study by Barrajas et al., they investigated 30 lesions in 27 patients treated with SRS and determined that both the mean and maximum rCBV was higher in recurrent tumor, and using an rCBV cutoff of 1.52, their results yielded a sensitivity of 91% and specificity of 72%.⁴⁴However, they noted a large degree of overlap in rCBV values and instead suggested the use of percentage of signal-intensity recovery (PSR), which they calculated by comparing the lowest signal-intensity post-contrast with the end post-contrast signal. Lower PSR values were found in the recurrent tumor group, and using a cutoff value of 76%, this yielded a sensitivity of 96% and specificity of 100%. Perfusion MRI is limited by tumors located near blood vessels, air, and bone and results in a longer examination time.

Figure 1.

(A) 62 year old female with a history of limited stage small cell lung cancer treated with chemoradiation, maintenance immunotherapy, and hippocampal avoidance PCI was found to have a 2.5 cm left temporal brain metastasis. She underwent gamma knife radiosurgery with 15 Gy in a single fraction. (B) Interval MRI 3 months following gamma knife radiosurgery showed interval decrease in size of the treated lesion. (C) MRI 8 months later showed interval increase in size of the heterogeneously enhancing lesion. (D) Perfusion MRI showed increased CBV corresponding to the lesion. She underwent resection and surgical pathology showed metastatic small cell carcinoma with necrosis.

Magnetic Resonance Spectroscopy

MRS provides information regarding the metabolic composition of tissue, with multiple published metabolite ratios proposed throughout the literature. For example, in the setting of an increased lipid/choline ratio, an increased lactate/creatine ratio, or a decreased choline/creatine ratio, RN is favored; if choline/creatine or choline/N-acetyl aspartate is increased, then TP is favored.^45–47 However, MRS is limited by both lesion size and location near blood vessels, air, bone.

Chemical Exchange Saturation Transfer

Chemical exchange saturation transfer (CEST) is a molecular imaging technique studying the tumor microenvironment concentration and exchange of mobile proteins and peptides. Benefits of CEST include its relatively fast acquisition time without the need for contrast injection. A viable tumor possesses increased protein and peptide content in comparison to RN, and thus CEST signal is higher in the setting of tumor recurrence.⁴⁸ An initial pilot study of 16 patients with brain metastases scanned with CEST found the best separation for RN and TP was obtained by using the magnetization transfer ratio (MTR) for CEST peaks corresponding to both the nuclear Overhauser effect (NOE) MTR and Amide MTR.⁴⁸ In a larger study of 70 patients with 75 brain metastases, several metrics showed significant differences in RN and TP. MTR Amide showed the best separation, with a specificity of 93% and sensitivity of 73% when using a 2 μT.⁴⁹

Positron Emission Tomography

Fluorodeoxyglucose PET (FDG PET) takes advantage of the cellular features of metastases, with viable tumors having increased radiotracer uptake while RN does not. However, because of differences in FDG PET methodologies and few studies using pathologic confirmation, in addition to the limitation of normal brain parenchyma demonstrating uptake, its usefulness in differentiating TP from RN is difficult to interpret with a wide range of reported outcomes (Figure 2).^50–52 Our institution investigated FDG PET and found a sensitivity of 75% and a specificity of 80%.⁵⁰ Other studies evaluating FDG PET have shown mixed results, where FDG was found to be non-specific or non-sensitive.^53,54A meta-analysis including 6 studies with FDG PET found a pooled sensitivity of 85% and sensitivity of 90%.⁵²

Figure 2.

(A) 60 year old male with ER+, PR+, HER2− breast cancer presented with a 3 cm left cerebellar brain metastasis. Patient underwent staged gamma knife radiosurgery, 30 Gy in 2 fractions, with treatments one month apart. (B) Initial follow-up imaging demonstrated tumor regression, but MRI brain 1 year later showed progressive expansile heterogenous enhancement. (C) Perfusion imaging demonstrated elevated rCBV along the superior aspect. (D) Brain positron emission tomography showed increased metabolic activity in the area of MRI enhancement, suggestive of tumor recurrence. He proceeded to resection with surgical pathology demonstrating radiation necrosis without viable tumor. Better diagnostic tools are needed as seen in this example.

In comparison to FDG PET, amino acid PET radiotracers demonstrate more selective mechanisms of amino acid uptake in viable tumors. Examples of amino acid radiotracers used in this setting include [11C]-methyl-L-methionine (C-MET), L-3,4-dihydroxy-6-[18F]-fluorophenylalanine (F-DOPA), O-(2-[18F]fluoroethyl)-L-tyrosine (F-FET), and 18F-fluciclovine.^55–60 Though no single amino acid radiotracer has been established as standard, the clinical usefulness of such radiotracers is apparent. In a study investigating 106 patients with glioblastoma or brain metastases with clinical suspicion of recurrence, the use of F-DOPA PET changed the diagnosis in 39% of patients and treatment plans in 17% of cases.⁵⁷ However, the use of many amino acid radiotracers is limited by the need for an on-site cyclotron for its development.

Fluciclovine or axumin is a readily available radiotracer used in the setting of biochemically recurrent prostate cancer but may also be used in the setting of brain metastases. Fluciclovine was investigated in a prospective pilot study of 20 lesions at our institution and found that SUVmax (maximum standardized uptake value) could accurately differentiate RN from TP, with a cutoff value of 4.3 resulting in a sensitivity of 100% and specificity of 63% to rule out progression.⁶¹ Another small study of 15 lesions also found fluciclovine to be accurate at different time points, with an SUVmax of 1.3 producing an accuracy of 100% and 87% at 30 and 55 min, respectively.⁶² The Study to Establish Image Interpretation Criteria for 18F Fluciclovine PET in Detecting Recurrent Brain Metastases (PURSUE; NCT04410367) and Study to Establish the Diagnostic Performance of 18F Fluciclovine PET in Detecting Recurrent Brain Metastases (REVELATE; NCT04410133) are looking to validate these results in a larger cohort.^63,64 Other nuclear medicine modalities include single photon emission tomography (SPECT) using thallium-201 and technetium-99. For now, no single modality is standard in this setting and remains investigational.

Treatment

The treatment of RN is often empiric, due to the invasiveness of pathologic confirmation. Management options include observation, medical therapy, or invasive procedures, and decision-making is largely driven by the acuity and severity of symptoms.

Observation

Observation with close interval follow-up and imaging can be considered in small asymptomatic lesions located ideally in noneloquent areas and whose natural course may be self-limited in nature. Volumetric increases of at least 20% of the original lesion size as measured by MRI can occur in as many as 32% of lesions treated with SRS and generally occur in the timeframe of 3–21 months post-SRS.³³ However, lesion enlargement may not be clinically apparent and can resolve over time. For example, Wang et al. observed that among 124 patients who received radiation treatment for nasopharyngeal carcinoma and developed temporal lobe injury, there was regression or resolution of white matter lesions, contrast-enhanced lesions, and cysts in 28%, 39%, and 7% of cases, respectively, on surveillance MR imaging.⁶⁵

Corticosteroids

Oral corticosteroids are considered first-line in the management of symptomatic radiation necrosis. There is data to suggest that the mechanisms underlying radiation necrosis are at least partly attributable to leaky angiogenesis contributing to vasogenic edema and an inflammatory cascade involving cell infiltrates and pro-inflammatory cytokines.⁶⁶ Consequently, oral corticosteroids lead to prompt symptomatic relief by reducing cerebral edema and modulating inflammation.⁶⁷ This benefit must be weighed against short- and long-term sequelae of steroid use (hyperglycemia, insomnia, psychosis, weight gain, Cushing syndrome, etc.). The lowest dose of corticosteroids to control symptoms is preferable. In a review of 169 patients with nasopharyngeal cancer who developed RN after radiotherapy and received either high-dose or low-dose intravenous methylprednisone, Zhuo et al. found no differences in clinical symptoms, cognitive functions, or treatment response on MRI.⁶⁸

Pentoxifylline/Vitamin E

The combination of pentoxifylline and oral vitamin E is an option for the medical management of mildly symptomatic RN or high-risk asymptomatic lesions that are large or close to eloquent areas. Pentoxifylline is a methyxanthine derivative thought to improve microcirculation by modulating blood viscosity while vitamin E is an antioxidant present in cell membranes, and the efficacy of this regimen has been primarily demonstrated in the prevention of skin fibrosis after breast irradiation.^69,70 Data as it pertains to RN are lacking, however. A pilot study by Williamson et al. of 11 patients with adverse radiation effects after SRS treated with 400 mg twice daily pentoxifylline and 400 IU twice daily oral vitamin E reported an average and maximum reduction of 72.3 mL and 324.2 mL, respectively.⁷¹ In this series, only one patient had enlargement, which was ultimately due to recurrence. Two patients had intolerable abdominal discomfort and nausea.

Bevacizumab

For patients with RN refractory to steroids, bevacizumab can be considered for escalation of medical therapy. Bevacizumab is an intravenously administered humanized monoclonal antibody targeting VEGF, a key mediator of angiogenesis implicated in radiation necrosis.⁶⁶ Commonly used regimens include 7.5 mg/kg every 3 weeks or 5 mg/kg every 2 weeks.^72,73 In a pooled analysis of 16 case series yielding 71 patients receiving bevacizumab for radiation necrosis following fractionated radiotherapy or SRS, Tye et al. reported radiographic improvement in 97% of patients with median decreases of 63% and 59% of T1 contrast enhancement and T2/FLAIR change, respectively.⁷⁴ Clinically, performance status improved in 79% of patients, and dexamethasone dosage was decreased by a median of 6 mg. A small clinical trial of 14 patients randomizing patients to bevacizumab at a dose of 7.5 mg/kg every three weeks versus placebo showed improvement in both neurologic clinical status and measured T1/T2 MRI volumes in all 5 patients initially assigned to bevacizumab and 7 patients initially assigned to placebo but crossed over to bevacizumab.⁷² A more recent clinical trial randomized patients with radiation necrosis after radiation for nasopharyngeal cancer to bevacizumab at a dose of 5 mg/kg every 2 weeks or corticosteroids.⁷³ Compared to corticosteroids, bevacizumab demonstrated improved mean percentage volume decreases on T1- (25.5% vs 5%) and T2-weighted (51.8% and 19.3%) MRI as well as recovery of clinical status (62.1% vs 42.6%). Although not absolute contraindications, patients at increased risk for bleeding or gastrointestinal perforations should consider alternative interventions.

Hyperbaric Oxygen

Hyperbaric oxygen therapy (HBOT) is another option for patients with asymptomatic or symptomatic RN by increasing perfusion and inducing angiogenesis. Patients undergo 20–40 daily treatments of hyperbaric oxygen at 2–2.4 atmospheres in a specially designed facility for 90–120 minutes at a time. Evidence supporting HBOT is largely limited to retrospective series and case reports, and response rates in these small series range from 80% to 90%.^75,76 The risk of seizures, ear barotrauma, sinus barotrauma, and pulmonary barotrauma are concerns with HBOT, and the availability of facilities and patient time commitment limit its usage.

Resection

Surgical resection is an invasive procedure reserved for symptomatic RN that is medically refractory or causing urgent neurologic compromise. Removal of the nidus of necrosis alleviates peri-lesional edema and mass effect, thereby relieving symptoms. A retrospective review conducted by Newman et al. of 46 patients with radiation necrosis predominantly after SRS with or without whole brain radiation noted that surgical resection decreased mean T2/FLAIR volume by 78% at 6 months, and greater extent of resection correlated with improvement of T2-FLAIR volume.⁷⁷ However, it is unclear if surgery leads to improved clinical outcomes since complications and neurologic deterioration could be as high as 54%, suggesting that patient selection is key.⁷⁸ Surgical resection also provides histologic confirmation in clinical scenarios where the diagnosis of RN versus TP is unclear.

Laser Interstitial Thermal Therapy

Laser interstitial thermal therapy (LITT) is a relatively novel invasive modality that serves as an alternative to surgical resection. LITT utilizes ablative hyperthermia generated by laser electromagnetic radiation to induce coagulative necrosis and abort the mechanisms underlying RN. LITT is often paired with a biopsy when the diagnosis remains uncertain. In a multicenter prospective study, Ahluwalia et al. reported on 19 patients with RN after SRS and found local progression-free survival of 100% at 12 weeks with a complete response by RANO criteria identified in all 4 patients who underwent total ablation.⁷⁹ Sujijantarat et al. noted excellent responses with LITT compared to bevacizumab in a retrospective review of 38 patients, where LITT correlated with improved overall survival and lesion volume reduction at 1 year.⁸⁰ Similarly, a meta-analysis performed by Palmisciano et al. comparing 148 patients undergoing LITT and 143 patients receiving bevacizumab demonstrated that LITT provided improved overall survival rates and equivalent symptomatic improvement and mean T1-weighted volume reduction.⁸¹ Similar to resection, LITT may not be feasible for tumors located in eloquent areas or in patients with poor performance status.

Conclusion

Radiation necrosis remains a relevant, unwanted complication following SRS for brain metastases. Risk factors such as radiation dose, radiation volume, use of systemic agents, and histology should be considered when counseling patients on their risk of developing RN. There is currently no single imaging modality that can definitively distinguish RN from TP, and a multi-modal approach should be taken to make an accurate diagnosis. Initial workup should include short interval MRI with perfusion studies, and if uncertainty remains, advanced imaging such as amino acid PET should be considered. Several treatment options exist for the management of symptomatic RN, with corticosteroids being first-line treatment.

The International Stereotactic Radiosurgery Society (ISRS) recently published its guidelines on the management of radiation necrosis. Recommendations from this guideline include the use of corticosteroids, bevacizumab, LITT, surgical resection, and HBOT as treatment options for radiation necrosis. Treatment recommendations are dependent on ISRS RN grade. Grade 1 is asymptomatic RN with no prior corticosteroid use, grade 2 is symptomatic RN with no prior corticosteroid use, grade 3 is symptomatic RN and steroid refractory, and grade 4 is symptomatic RN with neurological impairment and progressive necrosis despite trials of noninvasive treatments. For ISRS grade 1 RN, the panel recommends close surveillance with consideration of corticosteroids, while for grade 2 RN corticosteroids are recommended. In grade 3 steroid refractory RN, bevacizumab, per the guidelines, has the most evidence, but LITT, surgery, and HBOT may also be considered. In grade 4 RN with neurologic impairment and progression despite noninvasive treatment, surgical resection is recommended. It is important to highlight that these are guidelines based on expert recommendations, but treatment decisions must still be assessed in a multidisciplinary manner.

One of the major challenges regarding the interpretation and implementation of literature addressing RN is the heterogeneity amongst reports. This includes inconsistencies in anatomic imaging criteria and terminology (radiation necrosis vs adverse radiation effect vs pseudoprogression), as well as inconsistent reporting of SRS and concurrent systemic therapy variables. As we move forward, reports investigating RN should include similar terminology and variables to help clinicians make better conclusions. Future studies to create predictive models that can identify patients at highest risk of RN and noninvasive techniques that can accurately differentiate RN from TP are needed. This may allow for reduction of morbidity through earlier treatment and management.

Funding

None declared.

Supplement sponsorship

This article appears as part of the supplement “Pushing the Boundaries of Radiation Technology for the Central Nervous System,” sponsored by Varian Medical Systems.

Conflict of interest statement

S.T.C. (Research support from Blue Earth Diagnostics and Honorarium from Varian Medical Systems). J.H.S. (Scientific advisory board of NovoCure, EmpNia, and Neutron Therapeutics). S.S.L. (Kuni Foundation, research funding, Hutchinson Center as Lead Academic Participating Site; UG1 CA 233328; Japanese Society for Radiation Oncology, travel expenses; American College of Radiology, Alternate councilor on behalf of American Radium Society and Chair of CARROS Nominating Committee; Radiosurgery Society, Board of Directors and National Medical Director of the Distinction in Practice in Stereotactic Radiotherapy Program).

Authorship statement

The first draft of the manuscript was written by Z.S.M. and C.B. All authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Data availability

Not applicable.