Article Text

Download PDFPDF

Reversible brain dysfunction in MELAS: MEG, and 1H MRS analysis


This case report describes a follow up investigation of a patient with impaired word discrimination due to mitochondrial encephalopathy with lactic acidosis and stroke-like syndrome (MELAS) using proton magnetic resonance spectroscopy (1H MRS) and auditory evoked magnetic fields (AEFs). The initial 1H MRS showed no N-acetyl aspartate (NAA) and marked accumulation of lactate (Lac) in the stroke-like lesion of MELAS, which was silent in neural activity according to AEFs. The follow up investigations, however, demonstrated that NAA reappeared, that the formerly increased Lac signal was significantly reduced, and that the magnitude of AEFs of the lesion was markedly increased. Metabolic and functional changes in 1H MRS and AEFs reflected the neurological recovery very well. The stroke-like lesion was shown, using AEFs and 1H MRS, to be able to function properly, although brain tissue of the lesion initially had severe damage due to mitochondrial dysfunction.

  • magnetic resonance spectroscopy
  • magnetoencephalography

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.

Mitochondrial encephalopathy with lactic acidosis and stroke-like syndrome (MELAS) is characterised by recurrent stroke-like episodes, headache, seizures, neurosensory hearing loss, and point mutations of mitochondrial DNA. Diffuse fibrillary gliosis with abundant reactive gemistocytes and focal evidence of ischaemic neuronal injury, which might be due to metabolic disturbances, are seen on light microscopy. Electron microscopic examination showed that bizarre enlarged mitochondria are markedly aggregated in smooth muscle cells and endothelium of the cerebral blood vessels. On the basis of these findings, the stroke-like episodes in MELAS are considered to result from defects in neuronal metabolism, as well as in cerebral vasculature.1 2Neuroradiologically, MELAS is characterised by disappearance of the stroke-like lesions and their reappearance in another region.3 4

Proton MR spectroscopy (1H MRS) is a new tool for studying abnormalities in brain biochemistry.5-7 Many authors have reported that in ischaemic stroke 1H MRS shows characteristic prolonged increases in lactic acid (Lac) and lack ofN-acetyl asparatate (NAA), indicating “neuronal loss”.7 8 Acute stroke-like lesions of MELAS show similar findings to cerebral infarction, and tend to be normalised within a few months after onset.4 9 10However, it is still unclear whether metabolically improved brain tissue is still able to function in MELAS.

Magnetoencephalography (MEG) reflects intracellular electric current flow in the brain, providing direct information on neural activity. Since it was used for functional brain mapping of the primary sensory cortex by measuring somatosensory evoked magnetic fields,11 12 the accuracy of MEG source localisation has become accepted for clinical use.

The present case report describes the follow up of a patient with impaired word discrimination due to stroke-like lesions in bilateral temporal areas, using 1H MRS and auditory evoked magnetic fields (AEFs). Metabolic and functional changes of a stroke-like lesion of MELAS were analyzed and are discussed in comparison with the neuroradiological findings and clinical symptoms.

Case report

A 32 year old right handed man had severe headache and difficulties in communicating with others and was admitted to the outpatient clinic of Hokkaido Neurosurgical Memorial Hospital on 3 March 1999. Neurological examination showed severe sensory dominant aphasia. T2 weighted MRI showed a left temporal lobe hyperintense lesion involving the superior temporal region and Heschl's gyrus (fig1 A). Angiography showed no abnormalities and the neuroradiological findings did not correspond to the vascular supply territory. The presence of mitochondrial mutations of mtDNA 3243 in blood samples and muscle biopsy showing ragged-red fibres confirmed the diagnosis of MELAS. As he became free of neurological symptoms within the next 2 months, he was discharged.

Figure 1

(A) T2 weighted MRI showing a hyperintense lesion in the left superior temporal region on the first admission. (B) Proton MR spectroscopy (1H MRS) of the stroke-like lesion in the right temporal region on the second admission, and (D) at 4 months after the second onset showing the dramatic recovery in NAA/creatine+phosphocreatine (Cr) intensity ratio and the decline in Lac concentration. Vertical scales are the same over all 1H MRS. Cho=choline containing compounds.(C and E) T2 weighted MRI showing a hyperintense lesion in the right superior temporal area and the regions of interest (white squared line) of 1H MRS. T2 weighted MRI of an axial section and T1 weighted MRI of a coronal section showing the estimated dipoles of right N1m (white square) and P2m (white circle) of AEFs on the second admission (B) and at 4 months (D) after the second onset. The estimated current dipoles of both components (N1m and P2m) were finally located in the right superior temporal region.

Four months after discharge, he was readmitted because of headache and a new onset of auditory hallucinations. Although he showed only mild sensorineural hearing loss (>15 dB) at frequencies of 8 to 12 kHz, word discrimination acuity was markedly impaired.T2 weighted MRI on the second admission demonstrated a hyperintense lesion in the right posterior temporal area involving the superior temporal lobe and Heschl's gyrus although the left temporal lesion that was involved in the first episode appeared normal (fig 1C).

The AEFs were measured with a sampling rate of 600 Hz by a MEG system (VectorView, Neuromag, Finland) on the second admission, at 2 months and 4 months after the onset. Tone bursts of 1 kHz with a sound pressure level of 85 dB and duration of 200 ms were presented separately to the left or right ear via air tubes and inserts. Equivalent current dipoles (ECDs) were calculated using a single current dipole model and only the ECDs with a minimum correlation value exceeding more than 0.90 between the measured and the calculated fields were accepted.

Brain MR examinations were performed using a whole body 1.5T MR system (Signa, HiSpeed, General Electric, Milwaukee). Single voxel1H MRS with point resolved spectroscopy was performed with an echo time of 270 ms and a repetition time of 1500 ms 1 day before the AEF measurements. Volumes of interest (25×25×26 cm3) were placed in the bilateral posterior temporal regions containing the primary auditory cortex.

On the second admission 1H MRS showed a large Lac peak (Lac was not detected in normal tissue) and no NAA signal in the lesion of the right temporal lobe (fig 1 B). A clear NAA peak was found in the left temporal region and the NAA/Cr ratio was 1.88, which was slightly lower than the mean ratio of normal controls (2.76 (SD 0.48), n = 10). The range between the mean ± 2 SD of each metabolite in normal brain tissue was determined as the normal limit. The unaffected left parietal region thar seemed normal on MRI showed a normal NAA/Cr ratio of 2.38. Mild accumulation of Lac, however, occurred in both the left parietal and the left temporal regions.

The AEF responses in the right hemisphere were too small in amplitude to identify peaks. In the contralateral left AEF responses, N1m and P2m peaks were identified at 83 ms and 178 ms, respectively, after their stimuli and ECDs were localised close together in the supratemporal cortex (fig 1 C).

The patient showed dramatic recovery from impairment of word discrimination with no audiographic changes in the next 2 months.T2 weighted imaging showed that the hyperintensity diminished in the right temporal lesion. The right temporal NAA/Cr ratio was increased to 1.38 and Lac was markedly decreased to 0.91 of the Lac/Cr ratio. The right AEF responses were demonstrated as clear N1m and P2m at 95 ms and 182 ms, respectively, and the ECD of P2m was located in the superior temporal region. The ECD of N1m was, however, localised in the middle temporal white matter, which was remote from the auditory cortex.

A follow up MEG-MRI examination 4 months after the second onset showed the patient to be almost free from the neurological symptoms except for mild sensorineural hearing loss with no audiographic changes.T2 weighted MRI detected no abnormal intensity in bilateral temporal regions. A low NAA/Cr ratio (1.42) and a small peak of Lac (Lac/Cr ratio 0.63) were still persistent, but gradually normalised (fig 1 D). The right AEF responses showed a clear N1m peak at 86 ms and a P2m peak at 182 ms. There were no obvious differences from the left AEF responses. The ECDs of the right and left N1m and P2m were localised in the auditory cortices (fig 1 E).


This patient presented twice with acute cerebral focal mitochondrial decompensation and had auditory hallucinations and difficulties in word discrimination due to the bilateral temporal lobe lesions. The initial 1H MRS suggested that the lesion produced severe anatomical as well as metabolic alterations, and gave the impression that the tissue would hardly survive and regain active metabolic and functional states.

N-acetyl asparatate is synthesised only in the mitochondria of neurons13 and it has been considered a neuronal marker.8 14 As neurons in the CNS have an extremely limited capacity for regeneration, a decrease in NAA is usually interpreted as a sign of irreversible neuronal loss. A few previous 1H MRS studies of MELAS, however, reported that decreased NAA signals could recover, as in our patient, and speculated that mitochondrial dysfunction that caused the NAA decrease was reversible.4 9 Although the role of NAA is still unclear, a few authors have shown that the NAA concentration could predict clinical prognosis of patients with stroke and reflect neurological symptoms.6 15 16 Metabolic recovery of 1H MRS in our patient was well correlated with functional recovery of AEFs as well as clinical symptoms. The findings suggested that the residual NAA signal in the lesions indicates surviving viable neurons, which produce NAA and which can remain functioning. The NAA signal might, therefore, reflect neuronal function as well as the numbers of damaged but still surviving neurons.

We found increased amplitudes of N1m and P2m responses from the right hemisphere during the recovery period. As the P2m recovered before the N1m, later responses might revert to normal faster than early responses. Although there have been no studies of MEG signal recordings in patients with MELAS, a few reports described relations between evoked magnetic fields and cerebral infarction. Rossiniet al investigated N20m and P30m responses from the sensory cortical hand area after a monohemispheric lesion. They demonstrated independent changes of N20m and P30m, where the later response (P30m) became excessively larger in the affected hemisphere. They speculated that the later response tends to be preserved as it might be strongly related to more synaptic relays than earlier responses.17 Toyoda et alexamined AEF responses of P50m and N100m in cases of ischaemic stroke, and reported that earlier responses were predominantly affected.18 These findings, including the results presented here, suggest that independent sources contribute to earlier and later responses and stress that the later responses could be a good index for detecting initial signs of neurological recovery.

The most striking finding in the present study was that the stroke-like lesion was able to function properly within a few months after onset using MEG and 1H MRS, although the brain tissue involved in the lesion initially had severe damage due to mitochondrial dysfunction. Stroke-like lesions are difficult to treat and have different prognoses in individual patients with MELAS. However, we stress that some of these stroke-like lesions might show marked improvment. We think that this finding is encouraging for patients with untreatable stroke-like episodes of MELAS.


We thank Professor J Vieth for providing valuable comments on the manuscript.