Transient global amnesia (TGA) is a benign condition characterised by sudden anterograde amnesia of less than 24 hours without any further focal neurological deficit. It predominantly affects people older than 50 years and reoccurs in about 3–4% of cases only. In up to 10%, the amnesia may be accompanied by headaches, dizziness, nausea/vomiting, and neck pain.^1,2 Reported preceding factors range from emotional and physical stress, sexual intercourse, severe pains, sudden rapid temperature changes, and coughing attacks to iatrogenic triggering by cerebral angiography.^3–5 However, the underlying pathomechanism remains unclear. Several hypotheses, including arterial transient ischaemic attacks, epileptic seizures, hyperventilation, migraine, and spreading depression have been postulated and controversially discussed.^3,5,6 Recently, a transient mesiotemporal ischaemia induced by venous congestion was suggested as a further potential TGA cause.⁷ The postulated mechanism is thought to occur through a Valsalva-like manoeuvre triggered by some of the above mentioned preceding activities. Supporting evidence for this hypothesis was contributed by three ultrasound studies, which found a significantly higher prevalence of retrograde internal jugular vein (IJV) blood flow—attributed to IJV valve incompetence (IJVVI)—during a Valsalva manoeuvre (VM) in TGA patients compared with controls.^8,9,10 However, a mesiotemporal venous congestion implies a sufficiently patent venous pathway from the affected IJV through the sigmoid sinus, transverse sinus, confluens, the unpaired straight sinus (SS), and vein of Galen into the paired basal vein of Rosenthal and internal cerebral veins, which drain the mesiotemporal brain regions (fig 1). The SS collects blood from both sides but often delivers it predominantly unilateral into one transverse sinus. The present study analyses IJVVI in TGA patients in relation to the SS drainage side by means of extracranial duplex ultrasound and intracranial venous magnetic resonance angiography (MRA).

METHODS

Study group

Patients were prospectively enrolled into the study if they fulfilled the Hodges criteria for TGA.³ Controls were recruited from routine outpatients sent to our ultrasound laboratory if they were free of any relevant arterial atherosclerotic disease—that is, free of haemodynamically relevant extracranial or intracranial stenoses (grade of stenosis <50%). Exclusion criteria for either group were a decompensated congestive heart failure or a history of central venous IJV cannulation. All patients and controls gave informed consent.

Ultrasound protocol

Duplex ultrasound was performed in patients and controls in a head straight, supine position with a 7–11 MHz linear transducer (Powervision 6000, SSA-370A, Toshiba, Tokyo, Japan). IJV b-mode image and Doppler flow were continuously insonated approximately 2 cm above the internal jugular valves using the triplex mode of the ultrasound machine. The Doppler sample volume was set between 0.5 and 1 cm and placed in the centre of the IJV lumen. IJVVI was assessed on both sides after repeated training of the subjects during three maximal VMs (duration >5 s). Pictures of the VM were stored on magneto-optical disk. A sufficient VM was confirmed by a clearly visible increase of IJV diameter (approximately 1 cm or more) in the b-mode image. IJVVI was concluded if retrograde jugular flow could be seen in the Doppler spectrum during each VM (fig 2). Reduction of flow velocity or zero flow was rated as competent jugular valves.

MRI protocol

Intracranial non-contrast venous MRA was performed in patients only using 1.5 T scanners (Magnetom Vision or Magentom Symphony, Siemens, Erlangen, Germany). For data acquisition a two dimensional time-of-flight (TOF) MRA protocol was applied (TR 30 ms, TE 6.5 ms, FA 50°, 3 mm slice thickness, 1 excitation 7.19 min, 250 mm FOV, matrix 256 mm). Off-line analysis of the three dimensional MRA data sets was carried out subsequent to data transfer through the intranet on two workstations (SIENET MagicView VB33/Siemens, Germany, and Vitrea/Vital images, USA) using maximum intensity projections (MIP) of about 4 mm slab thickness orientated along the SS axis (fig 3). The SS, confluens, and adjacent horizontal parts of both TS were included in the slab volume. Appearance of the SS drainage side was evaluated by two investigators (RK, SJS) unaware of the presence or side of IJVVI and differentiated into three groups: a predominantly right sided (group A), a predominantly left sided (group B), and a bilateral drainage (group C, fig 3).

Statistical analysis

All presented values are given as means ± standard deviation. Two-sided Fisher’s exact test was performed for comparison of controls and patients. A p value <0.05 was considered significant.

RESULTS

Clinical findings

Twenty six patients were screened; one had to be excluded because of intracranial bilateral high grade stenoses within the carotis-t region. Twenty five patients (9 women, 16 men) aged 61 ± 10 years (range: 37–77 years) and 85 controls (37 women, 48 men) aged 62 ± 9 years (range 48–81 years) were included. The average duration of amnesia in patients was 7 ± 5 hours (range 20 minutes to 18 hours). Nine of 25 patients (36%) had a reliable history of a probable preceding VM (house cleaning and moving of furniture, football play, car mending, fitting a carpet, carrying of flower pots, digging in the garden, sawing wood, skiing, cough attack).

IJV valve incompetence

Ultrasound analysis for IJVVI was performed and evaluated in all controls and patients. IJVVI was found in 28 (33%) controls and in 17 (68%) of TGA patients yielding a significant difference (p  =  0.0025). Table 1 illustrates the distribution of right sided, left sided, and bilateral IJVVI, which did not differ between both groups. Two of the eight patients with competent valves (25%) and seven of the 17 with IJVVI (41%, p  =  0.66) had a preceding VM.

Venous MRA

MRA was performed in 20 of 25 TGA patients. Analysis of SS anatomy revealed a predominantly right sided drainage in eight (40%), a left sided drainage in five (25%), and a bilateral drainage in seven (35%) cases. Thirteen of the 20 patients did have a IJVVI and eight a preceding VM. The relation of SS drainage and IJVVI sides of patients with and without VM is illustrated in Table 2. IJVVI and SS drainage corresponded in five (63%) of the eight patients with VM and four (33%) of the 12 patients without a reported VM (p  =  0.0994).

DISCUSSION

The venous congestion hypothesis for TGA, originally postulated by Lewis in 1998,⁷ is primarily based on the theoretical consideration that the onset of symptoms often correlates with prior Valsalva-like activities, which may lead to venous reflux through the IJV. IJVVI may be caused by valve insufficiency or aplasia. In anatomical studies, aplasia has been reported to occur in up to 16% of non-selected patients whereas valve competence was not assessed.¹¹ Analysing IJV valve function, three US studies found a significantly higher prevalence of IJVVI in TGA patients, which supports the theory of a causal relation. Sander et al⁸ analysed the extent of duplex colour inversion during VM whereas Akkawi et al^9,10 used an echo-contrast method evaluating the number of refluxing contrast bubbles in the IJV. Both research groups reported an IJVVI prevalence between 70 and 73% in patients compared with 30 to 47% in controls.^8,9,10 Data of the current study (68% and 33%) confirm this phenomenon and also indirectly validate our chosen duplex US protocol. In addition, the distribution of the affected sides in patients and controls (right/left/bilateral: 68/32/0% and 59/35/6%) appears to be similar to data from Akkawi et al (60/3/37% and 89/0/11%).¹⁰ As the high prevalence of right sided IJVVI might correspond to the frequently observed right sided anatomical predominance of IJV drainage,^12–14 we additionally performed jugular blood volume flow measurements, which did not yield a correlation between IJV flow dominance and the side of the IJVVI (data not shown).

Assuming a causal and not coincidental relation of IJVVI and TGA, two further preconditions must be accomplished to effectively induce a mesiotemporal blood flow disturbance. These are a patent venous pathway and a plausible history of a VM. Concerning the intracranial venous anatomy, the unpaired SS, collecting blood from both mesiotemporal regions through the basal veins of Rosenthal and internal cerebral veins, is a key vessel as its drainage pattern into the confluens can vary considerably.^14,15 Interestingly, the correspondence of predominant SS drainage and IJVVI sides was slightly but not significantly higher in patients with a VM (63%) compared with those without (33%, p  =  0.0994). However, these findings still leave the TGA in 37% of cases with VM unexplained by the hypothesis. Finally, we could not see any further specific venous morphological characteristics in our TGA patients. Unfortunately, these findings do not deliver further evidence in favour of the venous congestion hypothesis. Yet, they also do not reliably exclude the postulated mechanism as some methodological limitations of our study have to be considered: venous MRA is an increasingly used technique for diagnosis or exclusion of sinus venous thrombosis.¹⁶ However, non-contrast MRA in regions with complex flow patterns might lead to MR signal loss and misinterpretation of venous anatomical variants.¹⁷ Furthermore, the side of the predominant SS drainage might not be as relevant as postulated because intravascular pressure changes can also be transmitted through a non-dominant small or hypoplastic but communicating vessel segment. Also, the postulated functional disturbance might not require a specific morphological correlate as various other pathways of the complex valveless intracranial venous network—for example, the vertebral veins or the paraspinal venous plexus could as well transmit the venous flow alteration. Finally, amnesia might be caused by functional disturbances of other than the mesiotemporal brain regions—for example, the thalamus,¹⁸ which comprise a different venous drainage pattern.

Regarding the second precondition of the venous congestion hypothesis, the triggering VM, only 36% of patients in our study population fulfilled this criterion. However, this percentage might be an underestimation as it mainly relies on statements of the patients, which could have suffered from a partial retrograde amnesia. In other reports the proportion of unusual TGA precipitating activities ranges from 30–81%.^1,4,8,10 However, some authors also included factors like ‘emotional stress’,¹⁰ which, in our opinion, do not reliably imply a likely VM. As all of the reported patient populations contain a proportion of ‘spontaneous’ TGAs the relevance of an initiating VM alone has still to be questioned.

In conclusion, the present study confirms the finding that TGA patients have a significantly higher prevalence of IJVVI. However, only approximately one third of them had a reliable history of a precipitating VM. We could not find any IJVVI related morphological intracranial venous drainage patterns on non-contrast venous MRA. Both findings argue against a simple causal relation between IJVVI and TGA. If VM and venous reflux do play a role within the aetiology of TGA, we assume that this probably occurs as cofactors of a still not understood multifactorial triggering mechanism, which could explain why reoccurrence rates of the condition are so low.

Research efforts to further analyse the relation between VM, IJVVI, and TGAs should be made towards an even more precise—for example, contrast-enhanced analysis—of the cerebral venous anatomy, including other venous drainage pathways like the vertebral and paraspinal veins in combination with a functional brain perfusion assessment. Assuming a multifactorial triggering mechanism, other coexisting contributing factors should be considered.