The pathophysiological events of the migraine aura and their relationship to the rest of the migraine syndrome persist as a challenging focus for research. This paper considers two non-invasive techniques that show a prospect of throwing light on these events: DC magnetometry and sequential magnetic resonance imaging (MRI). Each is capable of detecting and measuring different fundamental aspects of cellular pathophysiology. They are the first non-invasive techniques to have been used for detection of spreading depression in animals in vivo [1,2].
The spreading depression hypothesis, suggesting that the profound neurological symptoms in the migraine aura are attributable to phenomena akin to Leão's spreading depression , has a long history [4,5] and much recent support [6-10]. The chief rival hypothesis has generally been considered to be primary vascular insufficiency [11,12]. Spreading depression (SD) is a disorder that can be accompanied by both increases and decreases of perfusion . These are not the primary events, however, and SD can occur and propagate in a normal manner in avascular tissue . SD can be brought on by , or exacerbated by , vascular insufficiency, but this is not an essential factor in causing the substantial ionic disturbances  that render neurons inexcitable in SD.
The value of blood flow measurements in relation to SD is that they can be carried out with a variety of sophisticated non-invasive techniques, both single photon emission [6,8] and positron emission tomography . The hypoperfusion in both SD  and the migraine aura  outlasts the primary disturbance that causes short-lived neuronal dysfunction by an hour or more, and affects much larger regions of tissue at any one time. This makes the study of perfusion feasible and important, but it leaves a gap in the argument about the primary propagating disturbance. It is rather as if one were trying to identify an object travelling on a lake, by studying merely its wake. The wake left by the migraine aura seems similar to that of SD, but it might also be similar if the primary disturbance were vascular. Information is very much needed about the cellular events of migraine pathophysiology that are more directly linked to the disordered neural function.
Probably the clearest evidence for the occurrence of SD in migraine is still the very first evidence: the remarkable qualitative and quantitative similarity between the readily characterised symptoms of a typical visual aura [17,18] and the symptoms that would be predicted if SD, as studied in animals, were to occur in the primary visual cortex . We know from the clinical evidence that in at least some situations the disturbance of the migraine aura can propagate through seemingly normally functioning cortical tissue at a fairly constant rate of around 3mm/min, affecting each region transiently but profoundly for a few tens of seconds or minutes. The disturbance fails to propagate throughout the whole cortex, even on one side. Such propagation failure, often in sulci or at architectonic boundaries, is also characteristic of SD in primates , though in rat and rabbit SD tends to spread throughout most of the ipsilateral cortex . The propagation of the migraine aura can be halted with inhalation of CO2 and O2 gas mixtures , an observation that was at first taken to support the hypothesis that the disturbance was due to vasoconstriction , but which has since been shown to be true also for SD .
The cellular events that would most straightforwardly characterise the aura pathophysiology as due to SD are the very marked disturbances of extracellular and intracellular ion concentrations, for example the 10-fold increases of [K+]o and decreases of [Ca2+]o . As yet these are not amenable to direct non-invasive measurement. However, there are substantial electric currents that flow in consequence, associated with an extracellular negativity in the region of tissue most acutely affected . The distribution of currents changes only slowly as a wave of SD propagates through the tissue. In principle it might be detected non-invasively either by voltage measurements with electrodes on the skin (DC electroencephalography) or by sensitive magnetic measurements using non-invasive SQUID technology. In view of the attenuation of the voltage signals due to the skull and the very slowly changing nature of the signals there are in fact theoretical advantages to the use of magnetic measurements .
Magnetic signals during SD were first detected with an in vitro preparation, the turtle cerebellum . Subsequent in vivo experiments with anaesthetized rabbits  have revealed slowly changing magnetic signals, rising and falling over periods of around 10 minutes during propagating cortical SD . With repeated evocation of SD these were shown to bear a consistent relation to the events of SD, though the interference levels with DC magnetic studies  are such that it would have been difficult to be confident about the causation of such individual signals without repeated evocation and a suitable statistical approach . This leads to serious problems in the study of migraine patients, in whom the signals of interest may only occur once, if at all, in a recording session. Nevertheless, such challenging studies have been carried out , leading to reports of three types of magnetic signal that appear to be characteristic of migraine patients. These are sudden large amplitude waves, periods of reduced high frequency MEG activity, and slow DC shifts. None of these phenomena can presently be considered specifically indicative of SD. In particular, the changes of high frequency activity seem to this author more plausibly attributed to changes of arousal level in the migraine patients than to SD, since the fraction of neural tissue contributing to magnetic signals from human cortex that would be expected to be electrically silent at any one time on account of SD  would, on any hypothesis, be likely to be rather small. Similar fluctuations in high frequency activity were seen in the rabbit studies when anaesthesia was light, but not systematically related to the occurrence of SD . The slow DC shifts seen during the headache phase in three migraine attacks  do have a broad similarity to signals seen during SD in the animal experiments . Further elucidation of the duration, source and characteristics of such signals would however be required to establish their relationship to any electrophysiological events. In view of the acknowledged difficulties in handling potential sources of artefact in this kind of study, this is still a major challenge.
The issue of how to improve the rejection of artefacts in the measurement of magnetic signals with a rise and fall time of many minutes is interesting and has been addressed in recent work [25-27]. A technique has been developed in which a subject is constantly moved to and fro in a circular motion (25mm diameter with a period of 4sec) underneath a SQUID magnetometer. This means that steady or slowly changing magnetic sources in the patient generate signals at the frequency of the movement. These are much less susceptible to interference from magnetic sources, either outside the laboratory or on distant parts of the body, and to problems due to baseline shifts in the equipment . Automated data analysis extracts the components of the signals that are correlated with movement in two perpendicular directions. This technique can be used with standard multi-channel magnetometers positioned above the patient's head, yielding positional information about DC sources with essentially no baseline drift. As a test of performance, this technique has been used to measure the slowly changing magnetic signals due to changes of corneo-retinal current during light adaptation. This current rises and falls following light onset with a peak ca. 8min after onset . Using rotatory scanning for magnetic measurements in such an experiment, it is possible to follow the slow time course of the changes, which are quite obscured by interference using conventional DC magnetometry, even inside a high quality magnetically screened room [26,27]. This technique could improve greatly the amount of information obtainable from migraine patients, compared with simple DC magnetometry. It does however require dedicated (albeit relatively simple) equipment combined with clinical magnetometry facilities and patients with predictable aura.
The second technique for characterising cellular pathophysiological changes is Magnetic Resonance Imaging (MRI). It has recently become clear that MRI can be used with a number of different protocols to be selectively sensitive to different aspects of tissue physiology. A technique called gradient echo (or T2*) imaging was the first to be used to follow non-invasively the propagating disturbance of SD, in rat brain . Increases of MRI signal of up to 15% were seen locally for periods of 1-2min as a wave of SD propagated through the cerebral cortex. Several aspects of the very profound tissue changes occurring in SD could contribute to these changes in MRI signal. This particular protocol is sensitive to changes of tissue deoxyhaemoglobin content. Thus the changes may have been due in part to increases of blood flow and increased venous oxygenation in the acute phase of SD . MRI protocols used subsequently in similar experiments in a different laboratory  have indicated a decrease of the apparent diffusion coefficient for water occurring locally in the tissue during SD. It is unclear as yet to what extent distinct physiological parameters can be extracted cleanly with the help of combined MRI protocols. However, cell swelling, altered membrane permeability, restricted extracellular diffusion and changes of blood flow, blood volume and oxygenation are all capable of influencing MRI parameters. Each of these is known to change in some conditions during SD and all may eventually in principle be capable of disentanglement through clever application of MRI techniques. Much work is currently being devoted to this goal, particularly with in vitro MRI studies. As always in the study of the migraine aura however, one of the most challenging problems will be to succeed in studying patients during an attack. Since the equipment and expertise required for suitable MRI studies is available in many sites around the world, it is to be hoped that many opportunities will arise to generate significant data.
MRI has the merit of generating direct images of the sites of tissue affected in a disturbance. This, and the relatively simple technical considerations involved in making dynamic studies with suitable protocols once the equipment is available, make it an attractive technique for increasing knowledge of the cellular pathophysiology in the migraine aura. It cannot directly, however, reveal anything of the underlying electrophysiological changes. Magnetometry can in principle provide this, though with greatly limited capability for localising a distributed pathology such as SD. Magnetometry and MRI are substantially complementary in these respects, and a combined programme of study with advanced techniques in both fields should offer substantial hope of furthering the understanding of aura pathophysiology.
Parts of the author's work have been supported by the Wellcome Trust, The Migraine Trust, the Royal Society and the Medical Research Council.
1 Gardner-Medwin A.R., Tepley N., Barkley G.L., Moran J., Nagel-Leiby S., Simkins R.T. & Welch K.M.A. (1991). Magnetic fields associated with spreading depression in anaesthetized rabbits. Brain Res. 540,153-158
2 Gardner-Medwin A.R., van Bruggen N., Williams S.R. & Ahier R.G. (1994). Magnetic Resonance Imaging of propagating waves of Spreading Depression in the anaesthetized rat. J. Cerebral Blood Flow & Metabolism 14:7-11
3 Leão A.A.P. (1944). Spreading depression of activity in the cerebral cortex. J. Neurophysiol. 7:359-390
4 Leão AAP & Morrison RS (1945). Propagation of spreading cortical depression. J. Neurophysiol. 8:33-45
5 Milner P.M. (1958). Note on a possible correspondence between the scotomas of migraine and spreading depression of Leão. EEG Clin. Neurophysiol. 10: 705
6 Olesen J., Larsen B. & Lauritzen M. (1981). Focal hyperemia followed by spreading oligemia and impaired activation of rCBF in classic migraine. Ann. Neurol. 9:344-352
7 Gardner-Medwin A.R. (1981). Possible roles of vertebrate neuroglia in potassium dynamics, spreading depression and migraine. J. exp. Biol. 95:111-127
8 Lauritzen M. & Olesen J. (1984). Regional blood flow during migraine attacks by Xe-133 inhalation and emission tomography. Brain 107:447-461
9 Lehmenkühler A., Grotemeyer K.-H. & Tegtmeier F., eds. (1993). Migraine: basic mechanisms and treatment. Urban & Schwarzenberg. Munchen. 602 pp.
10 Woods R.P., Iacobini M. & Mazziota J.C. (1994). Bilateral spreading cerebral hypoperfusion during spontaneous migraine headache. New England Journal of Medicine 331: 1689-1692
11 Wolff H.G. (1980). Headache and Other Pain. 4th ed. (ed. D.J. Dalessio). New York: Oxfourd University Press.
12 Olsen T.S., Friberg L., Lassen N.A. (1987). Ischemia may be the primary cause of the neurologic deficits in classic migraine. Arch. Neurol. 44:156-161
13 Hansen A.J. & Lauritzen M. (1984). The role of spreading depression in acute brain disorders. An. Acad. brasil. Cienc. 56(4): 457-479
14 Martins-Ferreira, H. & Oliveira-Castro G. (1966). Light scattering changes accompanying spreading depression in isolated retina. J. Neurophysiol. 29:715-726
15 Gardner-Medwin, A.R. & Mutch W.A.C. (1984). Experiments on Spreading Depression in relation to Migraine and Neurosurgery. An. Acad. brasil. Cienc. 56(4): 457-479
16 Nicholson C. & Kraig R.P. (1981). The behavior of extracellular ions during spreading depression. In: The application of ion-selective electrodes. Ed. Zeuthen T., Elsevier, Amsterdam, pp 217-238
17 Lashley K.S. (1941). Patterns of cerebral integration indicated by the scotomas of migraine. Archs. Neurol. Psychiat. Chicago 46:331-339
18 Grüsser, O.J. (1995). Migraine phosphenes and the retino-cortical magnification factor. Vision Res. 35: 1125-1134
19 Marshall W.H. (1959). Spreading cortical depression of Leão. Physiol. Rev. 39:239-279
20 Bures J., Buresová O. & Krivánek J. (1974). The Mechanisms and applications of Leão's spreading depression of electrical activity. Academia, Prague.
21 Gardner-Medwin A.R. (1994). Very slow electrophysiological signals in man: advantages of study by magnetic measurement. J.Physiol. 479:143P
22 Okada Y.C., Lauritzen, M. and Nicholson C. (1989) Magnetic field associated with spreading depression: a model for the detection of migraine. Brain Res. 442,185-190
23 Barkley, G.L., Moran J.E., Takanashi, Y., Tepley N. (1991). Techniques for DC magnetoencephalography. J. Clin. Neurophysiol. 8:189-199
24 Barkley G.L., Tepley N., Nagel-Leiby S., Moran J.E., Simkins D.O. & Welch, K.M.A. (1990). Magnetoencephalographic studies of migraine. Headache 30: 428-434
25 Gardner-Medwin A.R. (1992). Rotatory scanning for improved discrimination of very low frequency components of the magneto-encephalogram. pp. 881-885 in Biomagnetism '91: Clinical aspects. N. Hoke et al. eds., Elsevier.
26 Gardner-Medwin, A.R. (1993). MEG studies of slow CNS disturbances. pp 229-238 in Migraine: basic mechanisms and treatment., ed. Lehmenkühler A. et al., Urban & Schwarzenberg: Munchen, 602 pp.
27 Gardner-Medwin A.R., Elbert T. & Kowalik Z. (1993). Measurement of slowly changing magnetic sources in the brain. IEE Colloquium Digest 16(9):1-3
28 Arden G.B. & Kelsey J.H. (1962). Changes produced by light in the standing potential of the human eye. J.Physiol. 161: 189-204
29 Latour L.L., Hasegawa Y., Formato J.E., Fisher M. & Sotak H.
(1994). Spreading waves of decreased diffusion coefficient after
cortical stimulation in the rat brain. Magnetic Resonance in Medicine