Radiation Necrosis Workup

Updated: Jul 20, 2021
  • Author: Gaurav Gupta, MD, FAANS, FACS; Chief Editor: Stephen L Nelson, Jr, MD, PhD, FAACPDM, FAAN, FAAP, FANA  more...
  • Print

Imaging Studies

A fundamental problem in the diagnosis of radiation necrosis is that most imaging studies do not preclude the need for surgical brain biopsy or craniotomy for diagnosis. The typical appearance of brain radiation injury is similar to that of brain tumors, with a contrast-enhancing mass surrounded by edema and mass effect. [18]

With conventional MRI, CT scan, positron emission tomography with [18 F]-labeled fluorodeoxyglucose (PET-FDG), and thallium 201 spectroscopy (single-photon emission CT [SPECT]), differentiating radiation necrosis from the recurrent tumor is difficult. [19]  Recently many advanced imaging techniques like diffusion-weighted images (DWI), perfusion-weighted Images (PWI), and Magnetic resonance spectroscopy (MRS) have been used to differentiate radiation necrosis from pseudoprogression or progression of the tumor. Most of the research has been focused on recurrent astrocytoma. See the images below.

MRI of a patient with symptoms of gait unsteadines MRI of a patient with symptoms of gait unsteadiness 1 year after being diagnosed with a posterior fossa primitive neuroectodermal tumor (PNET). Treatment during the 1-year interval prior to this MRI consisted of surgical resection, craniospinal radiation of 2340 cGy, boost dose given to the posterior fossa for a total of 5500 cGy, chemotherapy (vincristine, cis-platinum, and cyclohexylchloroethylnitrosurea [CCNU]), and dexamethasone therapy.
Positron emission tomography with [18F]-labeled fl Positron emission tomography with [18F]-labeled fluorodeoxyglucose (PET-FDG) performed following the MRI of a patient with symptoms of gait unsteadiness 1 year after being diagnosed with a posterior fossa primitive neuroectodermal tumor (PNET). Treatment during the 1-year interval prior to these studies consisted of surgical resection, craniospinal radiation of 2340 cGy, boost dose given to the posterior fossa for a total of 5500 cGy, chemotherapy (vincristine, cis-platinum, and cyclohexylchloroethylnitrosurea [CCNU]), and dexamethasone therapy. PET-FDG demonstrates hypometabolism consistent with probable radiation necrosis. Four years later, the patient is stable and without evidence of tumor progression.

CT scan

CT scan is not helpful in making the diagnosis of RN.

It is most useful in the acute, clinical decline of a patient with a brain tumor to differentiate acute hemorrhage from increased intracranial pressure, obstructive hydrocephalus, or a herniation syndrome.

Conventional MRI

MRI signal changes in radiation necrosis cannot be differentiated from tumor-related changes.

In a study by Asao et al, diffusion-weighted MRI sequences of radiation necrosis were associated with marked and spotty hypointensity compared with recurrent tumors, with maximal apparent diffusion coefficient values in each lesion being smaller for recurrent tumors versus radiation necrosis. [20]

Dequesada et al noted that lesions containing radiation necrosis never displayed gyriform lesion/edema distribution, marginal enhancement, or solid enhancements.  [21]

Dequesada et al reported that the lesion quotient (which is the ratio of the nodule as seen on T2 imaging as compared to the total enhancing area on T1 imaging) was associated with a quotient of 0.6 or greater in all cases of the recurrent tumor and a quotient of 0.3 or less was seen in 4 of 5 cases of radiation necrosis. [21]

T1, T1 with gadolinium, T2, T2- fluid-attenuated inversion recovery (FLAIR), and proton density do not adequately enable differentiation of radiation necrosis from tumors.

Reddy el al concluded that enhancement patterns on MRI were just as accurate in predicting pathologic diagnosis as MR spectroscopy. [22]

Previously, radiation necrosis was believed to have greater peripheral than central enhancement with gadolinium. However, this peripheral enhancement pattern is not a consistent finding in radiation necrosis. Tumors also may display a greater peripheral than central enhancement.

MRI patterns that may signal but are not diagnostic for the possibility of radiation necrosis include the following:

  • Non-enhancing tumors prior to surgery that have enhancing foci that subsequently develop within and circumscribed to the tumor bed may represent necrosis rather than progression to the higher-grade tumor.

  • Enhancing lesions that develop some distance from a primary glioma but within the radiation field may be indicative of radiation necrosis.

  • The enhancing focus in the periventricular white matter, particularly within the corpus callosum or the top of the ventricles, may represent necrosis because the periventricular white matter is highly susceptible to radiation necrosis.

  • Consider radiation necrosis if a new MRI enhancing lesion has a soap-bubble or Swiss-cheese appearance.

Perfusion weighted images (PWI)

Dynamic susceptibility contrast (DSC), dynamic contrast-enhanced (DCE), and arterial spin labelling (ASL) are the three MRI perfusion imaging techniques used. DSC and DCE rely on the injection of intravenous contrast agent whereas the ASL uses magnetically labeled blood as an endogenous contrast media. [23]  DSC is the most employed technique with the best diagnostic performance among the three. Regional cerebral blood volume (rCBV) measurement in DSC demonstrates a significant elevation in the setting of tumor progression compared to radiation necrosis. Mean rCBV thresholds in the range of 0.9 to 2.15 and maximum rCBV between 1.49 and 3.10 have predicted tumor progression with more than 95% accuracy. [24]  Quantitative hemodynamic indices like the transfer constant (Ktrans ) and fractional plasma volume (Vp )  measured in DCE showed higher values in true progression compared to radiation necrosis [24, 25]  (Figures 2A and 2B).

Diffusion-weighted images (DWI)

Homogeneous or multifocal high signal intensity on diffusion-weighted images was observed in tumor progression compared to peripheral or no hyperintensities in cases of pseudoprogression and radiation necrosis. [24, 26]

Magnetic resonance spectroscopy

Magnetic resonance spectroscopy (MRS) offers a new, quantitative approach to help differentiate radiation necrosis from tumor recurrence.

A few studies with histologic confirmation demonstrate the potential of MRS in differentiating radiation necrosis from tumor recurrence.

Future studies will determine the usefulness of MRS in avoiding biopsy or craniotomy for definitive diagnosis. MRS measures various brain metabolic markers, as follows:

  • Creatine is indicative of cellular bioenergetics

  • Choline (Cho) compounds reflect membrane metabolism

  • Lactate (Lac) reflects anaerobic metabolism

  • N -acetyl aspartate (NAA) is an amino acid marker of neurons

  • Gliomas have high peaks in Cho and even higher peaks in creatine when compared to healthy brain tissue

  • Radiation necrosis has decreased peaks in Cho, NAA, and creatine compared to healthy brain tissue

  • MRS may be of particular use in distinguishing pure tumor from pure necrosis. Knowledge of the choline/creatine and lipid-lactate/choline ratios may allow one to distinguish between tumor and necrosis; the good correlation between MRS and lesion biopsy findings was reported from a study by Rock et al. [27]

The multimodal assessment using a combination of the above modalities may have a better predictive ability in differentiating progression from radiation necrosis. [24] A low apparent diffusion coefficient (ADC), high regional cerebral blood volume (rCBV), high transfer constant (Ktrans), high fractional plasma volume (Vp), and high choline: creatinine ratio in multiparametric MRI suggested tumor recurrence rather than radiation necrosis. [24, 28]

Dynamic testing

Dynamic testing (eg, PET-FDG, SPECT) detects differences in tissue metabolism.

Tumors have greater metabolism (ie, increased uptake of PET-FDG and SPECT) than healthy brain parenchyma and areas of radiation necrosis.

Radiation necrosis is hypometabolic (ie, decreased uptake of FDG and thallium) compared to healthy brain parenchyma.

PET-FDG uses glucose transport and glycolysis as markers of metabolic activity.

Other PET imaging tracers have been proposed, including fluoride-labeled boronophenylalanine and other amino-acid tracers. These may be more useful in the detection of tumors because the background protein metabolism activity is lower than sugar and the background activity of brain glucose metabolism may complicate interpretations of PET-FDG. [29]

13 N-NH3 PET may also be useful because this compound is associated with high uptake in even low-grade astrocytomas, thus possibly allowing clinicians to distinguish between recurrent astrocytoma and radiation necrosis. [30]

Thallium metabolic activity is due to its similarity to potassium.

Thallium SPECT reflects the metabolic activity of sodium/potassium ATP-dependent membrane transport, chloride transport, and calcium channels.

In dynamic testing, a region of interest (ROI) is compared to a similar area of a healthy brain. An ROI located in one hemisphere is compared to a similar area in the contralateral hemisphere.

For bihemispheric lesions, an ROI is compared to an equivalent region in the anterior-posterior or posterior-anterior areas of brain parenchyma.

Most medical centers use qualitative assessments of ROI rather than quantitative assessments.

Despite diagnostic benefits and limitations of dynamic testing, histology often demonstrates mixed findings of malignant cells and radiation necrosis.

Advantages of PET-FDG [31]

PET-FDG correlates with prognosis and survival for newly diagnosed astrocytoma. Astrocytoma research has demonstrated that increased FDG uptake correlates with decreased survival.

Increased FDG activity on PET is more indicative of higher-grade astrocytomas such as anaplastic astrocytoma or glioblastoma multiforme.

PET-FDG also has assisted in guiding brain biopsy sites. Since brain biopsies are subject to sampling error, PET-FDG can assist the surgeon in obtaining the most metabolically active tissue to allow more accurate tumor staging.

PET-FDG is diagnostically useful in evaluating a tumefactive lesion (ie, a structural lesion on MRI that suggests tumor) when clinical history suggests another diagnosis (eg, stroke, demyelination, abscess).

Despite the potential of this tool in a newly diagnosed brain tumor, PET-FDG becomes problematic when differentiating radiation necrosis from tumor recurrence.

Disadvantages of PET-FDG

Its sensitivity and specificity in differentiating radiation necrosis from tumor recurrence are related to various factors. Overall, the sensitivity of FDG-PET has been reported as 80-90%, but the specificity is lower (50-90% depending on the series). [7]

An ROI of less than 1.6 cm lowers sensitivity and specificity in the differentiation of radiation necrosis from tumor recurrence. [32]

An ROI located in the temporal lobes and brain stem may have poor resolution due to artifacts from nearby bony structures.

Inflammatory cells in areas of radiation necrosis may show increased metabolic activity, which falsely can indicate tumor recurrence.

Tumor cells also may be present in areas of low glucose activity on PET-FDG.

Brain tissue used for comparison to the ROI can become depressed metabolically from radiotherapy and/or chemotherapy.

Carbon C 11–methionine PET scanning may be a complementary study to FDG-PET. In one series, 31 of 35 brain tumors showed increased11 C-methionine despite isometabolism or hypometabolism on FDG-PET scans, and 10 benign lesions (of which 2 were cases of radiation necrosis) showed decreased or normal uptake of11 C-methionine.  [33]

Thallium single-photon emission CT scan

Except for being more readily available at more medical centers than PET-FDG, thallium SPECT has the same limitations in dynamic testing.

A thallium index greater than 1.5 generally correlates with anaplastic astrocytoma, glioblastoma multiforme, primary CNS lymphoma, or metastasis. Overall, imaging studies provide additional information, but they cannot provide a definitive diagnosis (ie, to avoid biopsy or craniotomy).

Overall, imaging studies provide additional information, but they cannot provide a definitive diagnosis (ie, to avoid biopsy or craniotomy).



The similarities of radiation necrosis and tumor recurrence in clinical presentation and diagnostic imaging make performing a brain biopsy critical for diagnosis. [18]

Diagnosing radiation necrosis is problematic. The diagnosis depends on obtaining adequate biopsy findings and is prone to sampling errors up to15%. [34]

A brain biopsy sample must be large enough to exclude tumor recurrence without causing clinically significant neurologic deficits. Areas to avoid include the deep central areas of the thalamus, the motor strip, occipital lobe, and speech centers.

There is preliminary evidence that indicates the promising role of liquid biopsy as a diagnostic alternative to intracranial biopsies. Quantitative assays of Annexin V-positive microvesicles that are secreted into the bloodstream by GBM can aid in differentiating between tumor progression and pseudoprogression. [35]  Similarly, the ratio of myeloid suppressor cell-derived biomarkers HLA-DR and vascular noninflammatory molecule 2 expressions on CD14+ monocytes, termed the DR-Vanin Index (DVI) has been shown to distinguish RN from tumor progression with adequate certainty. [36]


Histologic Findings

Radiation necrosis tissue samples demonstrate necrotic tissue without predominance of malignant cells.

Irradiated tumor may contain necrosis, which does not necessarily signify radiation necrosis.

Some biopsy findings of radiation necrosis show both malignant cells and radiation necrosis.

A hallmark of radiation necrosis is involvement of the white matter with demyelination and oligodendrocyte dropout.

In addition to necrotic tissue, biopsy findings of radiation necrosis may demonstrate thickened vessels with endothelial proliferation and/or hyalinization with fibrosis and moderate infiltration of lymphocytes and macrophages.