Transient global amnesia: Acute SPECT functional imaging and neuropsychological deficits

Five patients with transient global amnesia (TGA) were seen clinically and studied neuropsychologically and with SPECT scanning within 24 hours in four and one within 48 hours of the attack. The memory deficits were characteristic of TGA and the magnetic resonance brain scans were normal in all patients. The SPECT scans showed hypoperfusion in the frontal and thalamic regions in all patients. In addition biparietal hypoperfusion was seen in one, parieta occipital in two and unilateral temporal in two patients. Transcranial Doppler studies were performed in three of the five patients and revealed abnormalities of possible relevance in one. Two of the SPECT studies had normalised when repeated at approximately 2 months post ictus. These five patients represent relatively "pure" forms of TGA as they all had normal MRI scans which excluded other disease processes such as subclinical or silent strokes. The frontal hypoperfusion in all five patients is of interest given the role of the frontal lobes in memory processing. Transient global amnesia (TGA) remains a clinical diagnosis of unknown pathophysiology with no corroborative laboratory or radiological test. Despite over one thousand case reports in the literature, 1-9 neuropshyhological assessment at time of or immediately after the ictus is limited to less than a dozen cases. 1,3-5 Strict definition as recommended by the Oxford criteria was not used in the majority of studies. 1 Several functional scanning studies, mainly single photon emission computed tomography (SPECT) reports and one positron emission computed tomography (PET) study, report of thalamic and/or neocortical hypoperfusion with hippocampal, temporal, frontotemporal, thalamus, hemicortical and mesial temporal sites of decreased cerebral blood flow. 3,10-19 Functional imaging with SPECT and PET may be useful in elucidating the cause for the most popular proposed mechanisms which include cerebral ischaemia, seizures and migraine. To date 10 such studies have been published describing 13 patients. In these studies, magnetic resonance imaging (MRI) scans were not reported on, or not performed in six studies and were abnormal in a further two. In the remaining two studies MRI scans were performed 10 days and three months post ictus. 3, 10-19 This report describes five patients in the acute phase, all with normal MRI scans. To our knowledge, this study represents the largest case series at TGA patients with normal MRI scans but abnormal SPECT scans.

The importance oftranscranial Doppler ultrasou nd and HMPAOSPECT in functional imaging of the brain is emphasised.

Abstract
Five patients with transient global amnesia (TGA) were seen clinically and studied neuropsychologically and with SPECT scanning within 24 hours in four and one within 48 hours of the attack. The memory deficits were Gharacteristic of TGA and the magnetic resonance brain scans were normal in all patients. The SPECT scans showed hypoperfusion in the frontal and thalamic regions in alf patients. In addition biparietal hypoperfusion was seen in one, parieta occipital in two and unilateral temporal in two patients. Transcranial Doppler studies were performed in three of the five patients and revealed abnormalities of possible relevance in one. Two of the SPECT studies had normalised when repeated at approximately 2 months post ictus.
These five patients represent relatively "pure" forms of TGA as they 30 SAJOURNAL

Methods
The five patients were derived from the Durban Stroke Data Bank (DSDB), (n=600) with presumed stroke as the admitting diagnosis.They were evaluated according to the tailored protocol of the DSD B which incorporates a prospective, contemporary stroke investigative evaluation including duplex and transcranial Doppler,~or computerised tomography (Cf) brain scanning, prothrombotic screens cardiac evaluation including transoesophageal echo, angiography and SPECT scanning where appropriate. Special attention is givento the assessment of higher cortical function defidts (HCID) with a screening examination.
In brief this encompasses eight major groups of H CFD (aphasias, amnesias, apraxias, agnosias,aledas neglect syndromes, frontal lobe syndromes and miscellaneous group) with subcategories, allpredefined. Focussed neuropsychological testing isperformed in those patients with H CID abnormality.

Clinical
The Oxford definition ofTGA was used to define the clinicalsyndrome (fable f).
Neuropsychological examination was performed by a neuropsychologist in four patients and by the first author in patient 2.
In all patients, the focus of the assessment was on memory functioning.

SPECfscanning
Twenty Millicuries 99m Technetium -HMPAO 99 is administered intravenously. Ceretec with Technetium produces a lipophilic complex which crossesthe blood brain barrier and is retained in the brain , Table I: The Oxford TGA study  diagnostic criteria for Transient  Global Amnesia' Attacks must be witnessed by a capable, reliable observer who was present at time of attack.
Clear cut anterograde amnesia must be present during the attack.
No alteration in consciousness or loss of personal identity may be present.
No focal neurological symptoms or functionally relevant signs may be present.
Epileptic features must be absent.
Attacks must resolve within 24 hours.
People with recent head injury or known active epilepsy are excluded.
Loss of personal identity is a feature of psychogenic or hysterical amnesia. It does not occur in patients with permanent memory loss due to structural brain disease Korsakolfs syndrome no matter how profound the degree of retrograde amnesia.
allowing assessment of regional cerebral blood flow.Imaging commenced 2-3 minutes after the injection of the tracer. The patient is positioned as comfortably aspossible with the room slightly darkened and is encouraged to be as relaxed as possible. Scans are obtained for 20 seconds with , regular 6 degree rotation of the camera head around the patient, completing a 360 degree rotation with approximately 60 images obtained. This required cooperation from the patient for aperiod of about 30-40 minutes.A medium to high resolution collimatorwas used on the gamma camera. Reconstruction was performed with positioning of the patient in the orbitomeatalline, in the transverse,coronal andsaggital planes.Three dimensional reconstruction was also performed. Images were interpreted on the computer screen comparing right and left regions of interest. This was done both visually and semi quantitatively. The transaxial slice with the largest perfusion defect was identified and a circular region of interest traced over that area.These circular regions of interest (ROf) were also obtained for six 31 SA JOURNAL OF RADIOLOGY. September 1996 areas of the grey matter. These included the cerebellum, thalami, basal nuclei, frontal, parieto occipital and temporal regions. These areas and their mirror images on the homologous region of the other hemisphere were stored on a template and adjusted when required for aparticular scan.The ROl radioactive counts were compared to the mean count of the two cerebellar hemispheres and expressed asa percentage ofthe mean cerebellar count according to the formula (1-ROl/mean cerebellar count).

Transcranial Doppler
The basal cerebral vessels and all major branches of the circle of Willis were insonated via the temporal and sub occipital windows with the EME-Transcan transcranialDoppler using mounted 2 MHz probes. Flow velocities and pulsatility indices (Gosling) were measured with special attention to the identification of intracranial stenoses and emboli detection (15 minutes monitoring) .

Case reports
Patient I

Neuroradiological investigations
Tl andT2 weightedMRIbrainscanningwas performed 12 hours after the onset and was normal. Transcranial Doppler sonography revealed abnormally low basilar artery velocity ofl 0-18 cms/sec. SPECT brain scanning approximately 10 hours after presentation,  Figure 1).The percentage decrease ofROI greaterthan 15% with respect to the mean cerebellar count isnoted inTable II.Follow up 2 months post ictus revealed normal neurological and focussed neuropsychological testing, normal transcranial Doppler insonation of the basilar artery and the SPECT brain scan showed normal perfusion bilaterally.

Patient2
A financial advisor aged 53 years awoke normally one Sunday morning, read the newspapers with his wife, had sexual intercourse and immediately thereafter kept repeating himself and could not remember details of the preceding day. Some events over the preceding week were also defective according to his wife.Within three and a half hours he had recovered back to normal as judged by himself and his wife. Stroke risk factors were not present apart from personal stress and an occasional moderate alcohol intake. His past medical history was unremarkable. He was seen within 3 hours of the attack. The neurological examination was normal and neuropsychological examination performed. Routine stroke investigations including blood counts, lipogram, glucose, erythrocyte sedimentation rate, electrolytes electrocardiogram and chest radiograph were normal.
The screening neuropsychological assessment comprised a clinical memory evaluation and the Boston Naming Test.The assessment:6ndings are detailed in Table II

Neuroradiologica1 investigations
The Tl and T2 weighted MRI brain scan performed approximately 24 hours afterthe event was normal. Transcranial Doppler 4 hours after the onset showed was notable for non insonation of the right posterior cerebral artery. The left posterior cerebral artery was easily insonated and the basilar artery flow normal. SPECT scanning was performed within 12 hours and revealed a visually appreciated left parieto-occipital perfusion defect.The percentage decrease ofRO I greater than 15% with respect to the mean cerebellar count is noted inTable II. Follow up 2 months post ictus revealed a persistent animal specific naming defect but this had improved. The trans cranial Doppler and SPECTbrain scan were repeated at this time and were both normal.

Patient3
A 54 year old woman awoke one morning unable to remember either the month or year or why she was at the place they were on holiday. She could recite accurately her home address and personal names. Repetitive questioning as to what she and her husband were doing at the holiday venue followed for the next few hours. No stroke risk factors were present and her relevant past medical history was otherwise unremarkable.
Examination, both systemic and neurological was normal. Bedside autobiographical review made it clear that her memory defectstretched back severalweeks prior to the ictus. Routine stroke investigations including blood counts lipogram, glucose, erythrocyte sedimentation rate, electrolytes electrocardiogram and chest radiograph were normal. She normalised clinically over a 12 hour period.
Neuropsychological evaluation was performed within 6 hours. The neuropsychological tests administered were the Wechsler Memory Scale,Rey'sAuditory Verbal Learning Test, Luria's Neuropsychological Investigation expanded by the second author to include autobiographical and general semantic memory and a Famous Faces Test. Overall she presented preservation of immediate memory, anterograde amnesia for both verbal and nonverbal material and retrograde amnesia characterised by a difficulty in recalling autobiographical episodes. Her orientation and immediate memory was normal astested by the Digit Span TaskAnterograde deficits for verbal memory was severely impaired and displayed a typical lowered and fluctuating plateau associated with frontal lobe impairment. Recognition memory was poor and recall of information during assessment was intermixed with word repetitions. Recall of material presented in context was severely impaired. Visual memory was mildly impaired. The neuropsychological assessment :6ndings are noted in Table II.

Neuroradiological investigations
CT and MRT (Tl and T2) brain scans and duplex and transcranial sonographywere all normal. SPECT scanning done within 12 hours of the ictus revealed visually appreciated hypoperfusion defects in the left frontal and parietal regions. The percentage decrease ofROI greater than 15% with respect to the mean cerebellar count is noted in Table II

Neuroradiological investigations
The MRI brain scan and duplex Doppler sonography ofhis cervicocephalic vessels were normal. The SPECT scan performed 24 hours after presentation revealed to visual inspection a marked right hemisphere hypoperfusion including the right frontal lobe and the right thalamic region. The percentage decrease ofROI greater than 15%with respectto the mean cerebellar count is noted inTable II.

PatientS
A 48 year old man suddenly questioned the type of shirt he was wearing while sitting on the edge of his bed one morning soon after wakening. The details were derived from his wife as he did not recollect anything about the event. He kept asking where he was, did not know that he had bought a new car 2 weeks previously. He asked about the shirt he was wearing atleast 4 times and kept blinking his eyes, asked about the day of the week, where he was,asked questions about the rugbymagazines next to his bed that he had just been reading. His wife recalled him touching his forehead and saying he had a "funny headache".The repetitive questioning lasted for about 20 minutes but for the rest of the day he seemed to be "floating and not with it" and he felt "fuzzy". He was in a very stress-fu1situation at the time, setting up his own business venture which was not going according to plan. Two weeks prior to the event he was away for the weekend with his wife and recalled details accurately according to his wife. His past history was unremarkable with no previous illness or surgery and no cerebrovascular riskfactors He was seen clinically within 24 hours of the attack and general and neurological examination was normal. Routine stroke investigations including blood counts, lipogram, glucose, erythrocyte sedimentation rate,electrolytes electrocardiogram and chest radiograph were normal.
Neuropsychological testing was performed within 48 hours and tests included the Rey Osterreith Complex Figure

Neuroradiological investigations
The MRI brain scan and duplex Doppler sonography of his cervicocephalic vessels were normal. The SPECT scan performed 48 hours after presentation revealed marked visually appreciated left hemisphere hypoperfusion, including the left frontal lobe and hypoperfusion of both thalamic regions.The percentage decrease of ROl greater than 15%withrespecttothe mean cerebellar count isnoted inTable II.

Discussion
The five patients described are unique in that they presented with classical features ofTGA as required by the Oxford TGA study diagnostic criteria, had normal MRI scans and abnormal SPECT scans in the acute stage. The SPECT scans and neuropsychological testing were performed within 24 hours of onset of the first symptoms in 4 patients and within 48 hours in one. Neuropsychological assessment was consistent with TGA in which the pervasiveness of anterograde and retrograde amnestic dysfunction is rapid in onset and in striking contrast to the preservation of immediate memory and the relative sparing of other higher mental functions. The MRl brain scanning is the most sensitive anatomical scan to exclude small infarcts and might be regarded as a prerequisite to exclude small or previously clinically silent infarcts in such patients. TGA re.mains a clinical diagnosis and one of the requirements is the absence of other neurological deficits MR1 scanning is an accurate method to exclude subclinical deficits in this setting. If sensitive paraclinical and radiological testing is also normal, as in these five cases,a purer form ofTGA presentation may be studied. The studies performed to date using PEPO,ISorSPECfl,II.17and 133Xenon l9 ,have indicated a variety of sites with perfusion defects, both bilateral and unilateral, in the acute stage ofTGA - Table IlI.The patients under discussion, with perfusion deficits noted both frontal and thalamic regions in all patients, the parieto ocdpital regions in 2 and unilateral temporal regions in 2, isto our knowledge, the largest case series of functional scanning in acute TGA together with normal MR1 brain scans.Although a variety of lesions, including cerebral infarcts and hae.morrhages have been reported in the temporal lobes and more commonly the thalami on C'I' brain scans 2 0-22 ofTGA patients, itmay be argued that such cases no longer represent pure forms ofTGA. In these cases, a transient neurological deficit may have been present very briefly at the onset of ictus with brain scan para clinical evidence of a vascular lesion akin to the events described as eITS (cerebral infarcts with transient symptoms).
The perfusion deficit might reflect a diaschisis effect of primarily thalamic dysfunction or neocortical in origin. In the study of Ott et al of unilateral amnesic stroke", a particular vulnerability forTGA was postulated with dysfunction in the left amygdalohippocampus or diencephalon. Whatever the mechanism, a primary diencephalic insult see.rnsplausible based on clinical and functional scanning data. All five of our cases showed relative thalamic hypoperfusion. With regard to the aetiology ofTGA, the normal MRI scans performed within 24 hours of onset argue against a cerebral infarct and subclinical brain damage. A recently proposed hypothesis posits that cerebral hypoperfusion may be secondary to a neuronal hypometa bolism itself initiated by excitotoxic neurotransmitter release temporarily impairing memory function. I The finding of the SPECfhypoperfusion in neocortical regions is consistent with such an hypothesis, the normal MR1 scans marshalling evidence against cerebral ischaemia as the primary mechanism. A recent positron emission computed tomography study in acute TGA demonstrated matched flow metabolism depression more consistent with neuronal dysfunction such as due to a seizural or migraine spreading depression mechanism.'°3 5 SA JOURNAL OF RADIOLOGY· September 1996 Certain neuropsychological findings are worthy of further comment. The dysexecutive findings in patients 1,3 and 5 and probably also patient 4, are consistent with a disruption in the organisation and control of the learning, retention and retrieval process highlighted by Baron et al.]°T he confrontation naming difficulty in patient 2 has also been noted by Kritchevsky et al.24 The SPECT scan revealed aleftparietal perfusion defidtwhich may be the basis of this problem. In patients 3 and 5visual memory was not as severely affected as verbal memory. A similar finding has been made by Walsh" in some of his cases. Patient 5 was assessed 48 hours post TGA onset and in this case the discrepancy may reflect the earner recovery of non verbal memory described by Okado et aFG in their two cases. The neuropsychological assessment findings would seem to support this in that although the main features ofTGA were the same in allfivepatients, there were individual differences in the severityand nature in which the symptoms manifested in each case. Constraints on neuropsychological assessment include the rapid onset and brevity of attack. Distress and frequently marked repetitive questioning frequently preclude a lengthy wide rangingneuropsychologicalassessment Our study reports on individuals free of cerebrovascular disease with minor risk factors, no previous illness of note, neocortical hypoperfusion with no evidence for a stroke mechanism on MR!scanning. Recent studies with relatively large numbers ofTGA patients argued strongly against a thromboembolic aetiology of TGAPThe transcranial Doppler (fCD) of low velocity in the basilar artery in case 1 may have reflected a cerebral hypometabolism. Incase 2, the finding of an absent signal from the left posterior cerebral artery and low basilar artery velodty with subsequent normalisation are to page 36 Transient global arnnesiaacute SPEeT functional imaging and neuropsychological deficits frompage35 consistent with posterior circulation embolic mechanism residua. The TCD findings, though interesting do not permit conclusions, asthe findings are nonspecific. SPECT scanning does not permit pathophysiological conclusions asa distinction between flow reduction caused by cerebrovascular disease or flow reduction due to neuronal dysfunction cannot be made. Cortical flow reduction secondary to thalamic diaschisis itself due to a posterior circulation ischaemia is a possible explanation. The 5 patients represent relatively pure TGA cases both clinically and on investigation with varying cortical hypoperfusion on SPECT scanning which normalised at one month follow up in 2.
TGA may bea core syndrome with several sites of dysfunction within the neuronal network subserving memory possible." In addition, support for a neuronal dysfunction rather than vascular mechanism, possibly thalamic based with secondary and transient cortical hypoperfusion seems most plausible in the cases described. This was first suggested by Triliet in 1987 using Xenon inhalation in TGA.19The caseseriesdescribed, givessupport to such an hypothesis because of the normal MRl scans in all five patients with its proven sensitivity for detecting cerebral infarction. The transient relatively mild frontal lobe disturbance noted in 4 of the 5 patients hints at a secondary disturbance such as that seen with diaschisis. Recent neuropsychological, PET and functional MR! data 2 8-31 provide strong support for the prefrontal cortex in human working memory. Activation of this area has been noted in cognitive tasks that are thought to be involved in declarative episodic or working memory. 28The frontal hypoperfusion in our 5 patients is consistent with such findings and also supports the theory of Baron et alia of a prefrontal metabolic depression, itself secondary due to thalamic dysfunction.