Article Text

Download PDFPDF

Diffusion weighted imaging in radiation necrosis
  1. V Biousse1,
  2. N J Newman1,
  3. S B Hunter2,
  4. P A Hudgins3
  1. 1Departments of Neurology, Neurological Surgery, and Ophthalmology, Emory University School of Medicine, Atlanta, Georgia, USA
  2. 2Department of Pathology, Emory University School of Medicine
  3. 3Department of Radiology, Emory University School of Medicine
  1. Correspondence to:
 Dr Valérie Biousse, Neuro-Ophthalmology Unit, Emory Eye Center, 1365-B Clifton Rd, Atlanta, Georgia 30322, USA;


This case report suggests that magnetic resonance imaging with diffusion weighted imaging may help distinguish between tumour recurrence and radiation induced necrosis in patients previously treated for a brain tumour.

  • radiation necrosis
  • diffusion weighted imaging

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Although non-invasive imaging techniques such as positron emission tomography (PET), single photon emission computed tomography (SPECT), magnetic resonance imaging (MRI), spectroscopy MRI, and dynamic susceptibility contrast MRI have improved our ability to diagnose radiation necrosis, the definite diagnosis of this condition may be difficult and brain biopsy is often required.1,2 We report a case supporting the use of diffusion weighted MRI to differentiate radiation necrosis from tumour recurrence.


A 30 year old woman presented with a one week history of progressive right hemiparesis, ataxia, and diplopia. Her past medical history was notable for a low grade cerebellar pilocytic astrocytoma in the region of the vermis, which was surgically resected when she was 15 years old. Eight years later, she developed bilateral fourth nerve palsies. Repeat MRI revealed abnormal tissue at the superior aspect of the vermis. Suboccipital craniectomy and tumour debulking were repeated, and pathology confirmed recurrence of a pilocitic astrocytoma. The patient subsequently underwent external radiation therapy (total dose of 5400 R to the whole brain, the tumour bed, and the spinal axis) and recovered well. Yearly follow up MRI showed stable clinical and neuroimaging findings, consisting of postoperative changes in the posterior fossa from the previous midline craniotomy, and mild encephalomalacia in the medial cerebellar hemispheres.

At age 30 years old (seven years after the last recurrence), she developed a right hemiparesis over one week, associated with binocular horizontal diplopia. On examination, she had a right hemiparesis sparing the face, without sensory abnormality. It was associated with a bilateral cerebellar syndrome, worse on the right, and with left sixth and seventh cranial nerve palsies.

Repeat MRI showed a small amount of nodular soft tissue within the fourth ventricle, which was unchanged compared with her previous MRI examinations, consistent with postoperative changes. There was also a new oval area of signal abnormality (11 × 9 mm) in the left upper pons which was hypointense on T1 weighted imaging and hyperintense on T2 weighted imaging. This lesion did not enhance (fig 1).

Figure 1

Axial T2 weighted image showing a round hypersignal in the pons.

A round lesion (12 × 12 mm) in the medial right temporal lobe and a smaller lesion in the lateral aspect of the right cerebral peduncle were hyperintense on both T1 and T2 weighted images. Neither of these lesions enhanced. On the T2 weighted gradient echo, these two lesions were mildly hypointense, suggesting small areas of haemorrhage consistent with occult vascular malformations. The periventricular white matter appeared hyperintense on the T2 weighted image, consistent with the radiation therapy. PET with fluorodeoxyglucose (FDG) failed to show any area of increased tracer activity. The MRI was repeated with diffusion weighted images. The lesion in the left upper pons and the lesion in the medial right temporal lobe appeared hyperintense. On the apparent diffusion coefficient (ADC) map, the medial right temporal lobe lesion was slightly hyperintense, suggesting a T2 shine-through effect. The upper pons lesion was hypointense, suggesting restricted diffusion (fig 2). Magnetic resonance angiography of the cervical and intracranial vessels was normal. A coagulation work up and echocardiography were obtained and were normal.

Figure 2

(A) Diffusion weighted image showing the pontine lesion as a hypersignal. (B) The same lesion is hypointense on the ADC map, suggesting restricted diffusion.

The new symptomatic lesion in the left upper pons was believed inconsistent with a recurrence of the primary tumour or a radiation induced tumour. The DWI findings were suggestive of pontine infarction, but radiation necrosis could not be ruled out. Biopsy of the lesion showed no evidence of tumour; there were areas of necrosis and hyalinisation associated with thick wall blood vessels, consistent with the diagnosis of radiation necrosis.


The appearance of a new lesion seven years after radiation treatment for a primary brain tumour may represent recurrence of the tumour, radiation induced demyelination, radiation necrosis of the brain, ischaemic stroke secondary to radiation induced vasculopathy, or rarely a secondary tumour or occult vascular malformation. It can be difficult to distinguish accurately among these possibilities, especially as the clinical features of these pathological conditions and their CT and MRI appearances can be quite similar.1–3 Our patient's neuroimaging showed many of these late complications from radiation treatment.

Chronic radiation injury leads to a diffuse leucodystrophy with confluent regions of subcortical gliosis and demyelination. It also damages cerebral capillaries, rendering the blood–brain barrier abnormally permeable, which explains the enhancement sometimes seen on cerebral MRI in regions of radiation necrosis. This makes the distinction between radiation necrosis and tumour recurrence very difficult in some patients.1–3 Tumour recurrence was unlikely in our patient. Indeed, the pontine lesion was remote from the tumour's primary site, it appeared very bright on the T2 weighted image, it did not enhance on the MRI, there was no mass effect, and our patient's FDG-PET did not show increased uptake of the FDG in the pons. SPECT and PET studies (especially FDG-PET) are now widely used to evaluate the metabolic activities of brain tumours and are usually helpful in differentiating tumour recurrence from radiation necrosis. However, these techniques are costly and require the use of a radioactive tracer.1 Recently developed MRI techniques such as dynamic contrast enhanced MRI and proton magnetic resonance spectroscopy have also been used in patients with brain tumours. However, because of low spatial resolution, discrepant results have been reported.2

Although Le Bihan et al suggested more than eight years ago that tumour recurrence and radiation damage might be distinguished using diffusion weighted imaging,4 we have been able to find only two anecdotal cases of radiation necrosis in which diffusion weighted imaging was done.5,6 Le Bihan et al hypothesised that high perfusion would be shown in tumour recurrence, while radiation damage would cause low perfusion and apparently decreased diffusion; these were the findings in our patient and in the two previously reported cases.5,6

Pontine infarction was suggested by the radiological findings in our patient, and she underwent a workup looking for causes of cerebral infarction. Radiation induced vasculopathy may affect both the large arteries (such as the basilar artery) and the small perforating vessels, and may cause infarction of any size. Pathology showed evidence of necrosis at the level of the pons, which is non-specific and could result from either infarction or radiation necrosis. However, the presence of thickened blood vessels in addition to necrosis made the diagnosis of radiation induced necrosis more likely.

The appearance of radiation induced necrosis on diffusion weighted imaging is of great interest as this may represent a non-invasive inexpensive way to distinguish between tumour recurrence and radiation necrosis in patients previously treated for a primary brain tumour, and result in a reduced number of brain biopsies. The fact that radiation necrosis presents as restricted diffusion on diffusion weighted imaging is not surprising given the potential role of ischaemia in the occurrence of radiation necrosis.



  • Competing interests: none declared