Vision Anomalies in Migraine
Bao Ngoc Nguyen
Submitted in total fulfilment of the requirements of the degree of Doctor of Philosophy
September 2013
Department of Optometry and Vision Sciences
The University of Melbourne ii Abstract
The aim of this thesis was to investigate visual responses of people with migraine in between attacks, in order to infer possible mechanism/s underlying migraine pathophysiology. The proposed neural abnormality in migraine is cortical hyperexcitability. To measure neural activity in the visual cortex, previous studies have measured the cortical evoked response to patterned visual stimulation; however, these have yielded inconsistent results. This raises the possibility that mechanisms other than cortical hyperexcitability might contribute to vision anomalies in migraine.
In this thesis, people with migraine (with and without aura) were compared to non- headache control participants. By measuring the cortical and retinal evoked responses simultaneously, Experiment 1 aimed to determine whether abnormalities in the cortical evoked response could be explained by dysfunction occurring earlier in the visual pathway. People with migraine showed reduced cortical responses but normal retinal function, indicating that the cortical deficits were unlikely to be a result of retinal dysfunction.
The purpose of Experiment 2 was to consider whether loss of visual field sensitivity in people with migraine was related to abnormal cortical and/or retinal function. Performance was compared between two visits: less than and more than seven days after a migraine. Migraine sufferers showed repeatable deficits in the cortical evoked response that were not worse in the days immediately after migraine, implying stability of the cortical deficit. In contrast, Experiment 2 found further reductions in sensitivity in the days immediately after a migraine attack, confirming previous reports. Some individuals showed monocular and/or localised visual field defects that were worse after an attack, suggesting that these might be retinal effects of migraine.
iii
The final experiment explored the balance between inhibition and excitation, one of the competing theories underlying cortical hyperexcitability in migraine. Using a perceptual centre-surround task to assess inhibition and excitation simultaneously, Experiment 3 found increased suppression in people with migraine, indicating increased inhibition, for stimuli of lower centre contrast. The contrast-dependent changes in perception were associated with abnormal contrast gain, as demonstrated by increased cortical evoked responses at low contrast and decreased responses at high contrast. Thus, Experiment 3 provided further evidence for cortical hyperexcitability leading to abnormal visual perception in between migraine attacks. However, cortical hyperexcitability cannot account for all vision anomalies in migraine. Rather, there is also the potential for adverse sequelae of migraine, possibly outside of the brain, in otherwise healthy and asymptomatic individuals (Experiments 1 and 2).
This thesis has implications for the use of non-invasive, clinical tests of visual function in clinical and research settings. People with migraine, on average, showed abnormal visual responses, even when asymptomatic and tested more than a few days after a migraine. This result reinforces previous suggestions that regular migraine sufferers should be excluded from normative databases. Additionally, there was considerable inter-individual variability in the visual responses, demonstrating that only a subset of individuals with migraine fall outside the range of normal visual performance. Further investigation of these particular individuals is potentially more informative regarding the pathophysiological processes of this debilitating condition.
iv Declaration
This is to certify that:
i. the thesis comprises only my original work towards the PhD except where indicated in the Preface,
ii. due acknowledgement has been made in the text to all other material used,
iii. the thesis is fewer than 100 000 words in length, exclusive of tables, maps, bibliographies and appendices.
Bao Ngoc Nguyen
v vi Preface
Part of the work presented in Chapter 2 has been published in the following journal article:
Nguyen, B.N., McKendrick, A.M. and Vingrys, A.J. (2012). “Simultaneous retinal and cortical visually evoked electrophysiological responses in between migraine attacks.” Cephalalgia 32: 896-907. and has been published in abstract form in the following conference proceedings:
Nguyen, B.N., McKendrick, A.M. and Vingrys, A.J. (2011). “Simultaneous pattern electroretinogram and visual evoked response in migraine.” Investigative Ophthalmology and Visual Science 52: E-Abstract 3521.
Part of the work presented in Chapter 3 has been published in the following journal article:
Nguyen, B.N., Vingrys, A.J. and McKendrick, A.M. (2013). “The effect of duration post-migraine on visual electrophysiology and visual field performance in people with migraine.” Cephalalgia. Aug 22. [Epub ahead of print].
Part of the work presented in Chapter 4 has been published in abstract form in the following conference proceedings:
vii Nguyen, B.N., Vingrys, A.J. and McKendrick, A.M. (2012). “Increased perceptual surround suppression of moving stimuli in migraine with aura.” Clinical and Experimental Ophthalmology 40: 133.
As primary author of the aforementioned publications, I was primarily responsible for the planning, execution, and preparation of work for publication in collaboration with my two supervisors, Associate Professor Allison McKendrick and Professor Algis Vingrys who provided assistance in the experimental design, data analysis, and editing of manuscripts.
The programming of the perceptual task in Chapter 4 was conducted by Associate Professor Allison McKendrick. Ms Jia Jia Lek assisted in the optometric screening and electrophysiological testing of control participants in Chapter 4.
viii Acknowledgements
To my supervisors, Associate Professor Allison McKendrick and Professor Algis Vingrys – thank you for everything.
To the Faculty of Science at the University of Melbourne – thank you for providing financial support through the Elizabeth and Vernon Puzey Postgraduate Scholarship.
To the Department of Optometry and Vision Sciences – thank you for providing me with a home for many years.
To Associate Professor Andrew Metha, Dr Michael Pianta, and Dr Larry Abel – thank you for being part of my PhD Advisory Panel.
To Dr Bang Bui, Dr Anne Weymouth, Dr Josephine Battista, Dr Zheng He, Dr Phillip Bedggood – thank you for helping with the initial setup and sanity checking of equipment and test protocols.
To the members, past and present, of my two research laboratories, the Clinical Psychophysics Unit and the Visual Functions Laboratory – thank you for keeping me fit and fed.
To all the research students and participants who volunteered for the experiments – thank you for the many hours you sat through.
To my family – thank you for understanding my selfish needs.
And to my dearest Jonathan – thank you for the love and life away from work.
ix
x Table of Contents
Abstract...... iii Declaration...... v Preface ...... vii Acknowledgements...... ix Table of Contents ...... xi List of Figures...... xv List of Tables ...... xvii
1 | A Review of the Literature...... 1 1.1 Introduction ...... 1 1.2 Migraine ...... 1 1.2.1 Classification and symptoms ...... 2 1.2.2 Pathophysiology of the migraine attack...... 4 1.2.3 Genetics of migraine...... 10 1.3 Cortical hyperexcitability...... 11 1.3.1 Evidence for hyperexcitability from the visual cortex...... 11 1.3.2 Changes in cortical hyperexcitability ...... 25 1.3.3 Investigations of mechanisms of hyperexcitability...... 27 1.4 Potential for pre-cortical dysfunction ...... 38 1.4.1 Parallel pathways...... 39 1.4.2 Concurrent pre-cortical and cortical dysfunction...... 45 1.5 Potential for adverse sequelae of migraine...... 53 1.5.1 Cortical deficits ...... 53 1.5.2 Retinal deficits...... 55 1.6 Overview of thesis ...... 56 1.6.1 Experiment 1...... 57
xi 1.6.2 Experiment 2...... 58 1.6.3 Experiment 3...... 59
2 | Experiment One...... 61 2.1 Introduction...... 61 2.2 Methods...... 63 2.2.1 Participants...... 63 2.2.2 Pattern electrophysiology...... 69 2.2.3 Statistical analysis...... 83 2.3 Results...... 84 2.3.1 The pattern visual evoked response (cortical response)...... 84 2.3.2 The pattern electroretinogram (retinal response)...... 88 2.3.3 Comparison between steady-state and transient responses...... 92 2.3.4 Relationship with migraine characteristics ...... 93 2.4 Discussion ...... 94 2.4.1 Retinal electrophysiological responses in migraine...... 95 2.4.2 Cortical electrophysiological responses in migraine ...... 96 2.4.3 Possible mechanisms for reduced cortical responses ...... 99 2.4.4 Greater dysfunction revealed by steady-state measures ...... 101 2.5 Conclusions...... 104
3 | Experiment Two...... 105 3.1 Introduction...... 105 3.2 Methods...... 108 3.2.1 Participants...... 108 3.2.2 Pattern electrophysiology...... 112 3.2.3 Visual field tests...... 112 3.2.4 Visual field analysis ...... 117 3.2.5 Statistical analysis...... 120 3.3 Results...... 122 3.3.1 Changes in the electrophysiological response...... 122 3.3.2 Changes in visual field sensitivity...... 130 xii 3.3.3 Patterns of visual field loss...... 138 3.3.4 Relationship between visual function and migraine features...... 142 3.4 Discussion ...... 143 3.4.1 Generalised visual field changes in the days after migraine...... 144 3.4.2 Localised visual field changes in the days after migraine ...... 145 3.4.3 Electrophysiological changes in the days after migraine ...... 148 3.4.4 Anatomical locus of visual anomalies in migraine...... 152 3.5 Conclusions...... 152
4 | Experiment Three ...... 155 4.1 Introduction ...... 155 4.2 Methods ...... 163 4.2.1 Participants ...... 163 4.2.2 Perceptual centre-surround task...... 164 4.2.3 Pattern electrophysiology ...... 168 4.2.4 Statistical analysis...... 171 4.3 Results...... 172 4.3.1 Perceptual centre-surround suppression...... 172 4.3.2 The pattern visual evoked response (cortical response)...... 175 4.3.3 Relationships between measures...... 182 4.4 Discussion ...... 185 4.4.1 Perceptual surround suppression in migraine...... 186 4.4.2 Cortical contrast gain in migraine ...... 190 4.4.3 Relationship between perceptual and electrophysiological responses . 194 4.5 Conclusions...... 196
5 | General Discussion and Conclusions...... 199 5.1 Overview of findings ...... 199 5.2 Heterogeneity of findings ...... 202 5.2.1 Differences in migraine characteristics ...... 202 5.2.2 Methodological differences...... 204 5.3 Possible non-visual mechanisms...... 207
xiii 5.4 Potential usefulness of tests for clinical purposes...... 209 5.4.1 Differential diagnosis of migraine...... 209 5.4.2 Differential diagnosis of ocular disease ...... 209 5.5 Future work...... 211 5.5.1 Anatomical locus of vision anomalies...... 211 5.5.2 Time course of changes in migraine...... 213 5.5.3 Physiological correlates for behavioural anomalies...... 214 5.6 Conclusions...... 215
6 | References ...... 217
7 | Appendix A ...... 249 8 | Appendix B ...... 281 9 | Appendix C ...... 311
xiv List of Figures
Chapter 1 | A Review of the Literature Figure 1.1 – Progression of visual disturbance in migraine aura...... 5 Figure 1.2 – Altered neurovascular coupling in migraine ...... 16 Figure 1.3 – Deficient habituation of the visual evoked response in migraine ...... 18 Figure 1.4 – Altered contrast gain of the visual evoked response ...... 22 Figure 1.5 – Model of surround suppression in V1...... 36 Figure 1.6 – Functional segregation of the primate visual system...... 40 Figure 1.7 – Parallel processing of visual information in primate ...... 44 Figure 1.8 – The visual pathway...... 48 Figure 1.9 – Example visual field results in migraine ...... 50
Chapter 2 | Experiment One Figure 2.1 – Self-reported migraine characteristics...... 68 Figure 2.2 – Schematic of electrode placement...... 71 Figure 2.3 – Average transient PERG and PVER waveforms ...... 79 Figure 2.4 – Average steady-state PVER waveform and Fourier spectrum...... 81 Figure 2.5 – Group PVER amplitudes...... 87 Figure 2.6 – Group PERG amplitudes...... 89 Figure 2.7 – PVER versus PERG amplitude...... 91
Chapter 3 | Experiment Two Figure 3.1 – Days since last migraine ...... 111 Figure 3.2 – Group PERG amplitudes...... 123 Figure 3.3 – PERG amplitude discrepancy plots...... 124 Figure 3.4 – Group PVER amplitudes ...... 127 Figure 3.5 – PVER amplitude discrepancy plots visit...... 128
xv Figure 3.6 – Group global indices of generalised sensitivity...... 131 Figure 3.7 – Average/Mean Defect discrepancy plots ...... 132 Figure 3.8 – Group global indices of localised loss...... 134 Figure 3.9 – Pattern Defect/Loss Variance discrepancy plots...... 135 Figure 3.10 – Total number of abnormal visual field locations...... 137 Figure 3.11 – Test-retest variability in visual field performance...... 139 Figure 3.12 – Example visual field defects in people with migraine...... 141
Chapter 4 | Experiment Three Figure 4.1 – Simulated changes to saturating cortical contrast gain...... 161 Figure 4.2 – Simulated changes to supersaturating cortical contrast gain...... 162 Figure 4.3 – Schematic of centre-surround task ...... 165 Figure 4.4 – Example staircase result for centre-surround task ...... 167 Figure 4.5 – Schematic of the electrode placement...... 169 Figure 4.6 – Mean suppression ratios...... 174 Figure 4.7 – Group suppression ratios for 20% contrast centre...... 175 Figure 4.8 – Mean PVER amplitudes normalised to control group...... 176 Figure 4.9 – Individual PVER amplitudes at 0%, 18%, and 97% contrast ...... 177 Figure 4.10 – Mean PVER phase normalised to control group ...... 178 Figure 4.11 – Mean PVER amplitude normalised to maximum contrast...... 180 Figure 4.12 – Contrast at which response was maximal ...... 182 Figure 4.13 – Relationship between perceptual and PVER measures...... 183 Figure 4.14 – Effect of non-linear contrast gain on contrast discrimination ...... 193
Chapter 5 | General Discussion and Conclusions Figure 5.1 – Proposed pathophysiological processes in migraine ...... 201 Figure 5.2 – Comparison between Experiments 1 and 3 ...... 206
xvi List of Tables
Chapter 1 | A Review of the Literature Table 1.1 – Diagnostic criteria for migraine without aura...... 3 Table 1.2 – Diagnostic criteria for migraine with aura ...... 3
Chapter 2 | Experiment One Table 2.1 – Summary of PVER measures ...... 86 Table 2.2 – Summary of PERG measures...... 88 Table 2.3 – Relationship between PVER and migraine features...... 93 Table 2.4 – Relationship between PVER and migraine with aura...... 93
Chapter 3 | Experiment Two Table 3.1 – Self-reported migraine features ...... 110 Table 3.2 – Summary of PERG measures...... 125 Table 3.3 – Summary of PVER measures...... 129 Table 3.4 – PVER interhemispheric asymmetry ...... 129 Table 3.5 – Average Defect and Mean Defect ...... 133 Table 3.6 – Pattern Defect and Loss Variance...... 136 Table 3.7 – Relationship between PVER and visual field performance...... 142
Chapter 4 | Experiment Three Table 4.1 – Self-reported migraine features ...... 164 Table 4.2 – Relationship with migraine features ...... 184 Table 4.3 – Relationship between perceptual and electrophysiological measures .. 184
xvii
xviii 1 | Chapter One A review of the literature
1.1 Introduction
The first part of this chapter will describe the diagnosis and epidemiology of migraine and summarise our current understanding of the pathophysiological processes that underlie this debilitating condition. This is followed by a detailed review of the changes to visual function in between migraine attacks and the anatomical, physiological, and psychophysical evidence for these alterations. Finally, given the existing literature, the aims and hypotheses of the project will be laid out in an overview of the thesis.
1.2 Migraine
Migraine is a common, disabling primary headache disorder, affecting approximately 15% of people worldwide (Stovner et al. 2007). It is ranked in the top 20 most disabling conditions worldwide (World Health Organisation 2001) and is the most debilitating of all neurological disorders in terms of years lived with disability (Leonardi and Raggi 2013). It is estimated that 90% of migraine sufferers report significant impairment of activities of daily living (Lipton et al. 2001a). The direct cost of medications and access to healthcare, plus the indirect cost of absenteeism and loss of productivity, was recently estimated at 18.5 billion Euros per year in Europe (Gustavsson et al. 2011). Migraine therefore poses a significant public health problem, warranting research into its pathogenesis, symptoms, and possible treatment strategies.
1 1.2.1 Classification and symptoms
Migraine is an episodic condition that is recognised purely by its symptomatology. The International Headache Society has provided a set of criteria to differentiate migraine from a range of other headache disorders, most recently updated in 2013 (International Headache Society 2013). The two most common forms of migraine are ‘migraine without aura’ (MO, Table 1.1) and ‘migraine with aura’ (MA, Table 1.2), which are the focus of this thesis. Both of these subtypes are characterised by headache, with nausea and/or vomiting and sensitivity to light (photophobia) and sound (phonophobia). The latter subtype is distinguished by the presence of ‘aura’, which are transient neurological disturbances of sight, speech, or tingling or numbness of the face or body. Other less common forms of migraine will not be discussed, but are described in detail in the International Headache Society classification guidelines (International Headache Society 2013).
The symptoms of a migraine attack generally occur in sequential phases, referred to as the ‘ictal’ period. Up to 48 hours before an attack, people may experience premonitory symptoms such as irritability, food cravings, and difficulty concentrating (Blau 1980; Kelman 2004b). The neurological disturbances of aura then occur, typically lasting five to 60 minutes and subsiding before the onset of headache (Rasmussen and Olesen 1992; Russell and Olesen 1996; Kelman 2004a). For people who do not experience a migraine prodrome, aura can be the first signal that a headache will follow. After the headache, people may experience post-dromal symptoms, including lethargy and moodiness (Blau 1991). All of these phases, although not always present at every episode, constitute a migraine attack. Once the attack has subsided, the person is typically asymptomatic until the next migraine occurs. The time period in between attacks is known as the ‘interictal’ period.
2 Table 1.1 – Diagnostic criteria for migraine without aura (MO) (International Headache Society 2013).
A Headache lasting 4 to 72 hours (untreated, or unsuccessfully treated)
Headache has at least two of the following characteristics: (1) unilateral location B (2) pulsating quality (i.e. throbbing or varying with the heartbeat) (3) moderate or severe pain intensity (4) aggravation by, or causing avoidance of, routine physical activity (e.g. walking or climbing stairs)
During the headache, at least one of the following additional symptoms: C (1) nausea and/or vomiting (2) photophobia and phonophobia
D At least 5 attacks fulfilling the above criteria
E Not attributed to any other disorder.
Table 1.2 – Diagnostic criteria for migraine with aura (MA) (International Headache Society 2013).
A Aura does not consist of motor weakness
Aura consists of at least one of the following fully reversible symptoms: (1) visual disturbances (e.g. flickering lights, spots, lines that may gradually spread across visual B field, blurring of vision, loss of vision) (2) sensory disturbances (pins and needles, numbness) (3) speech disturbances
Aura has at least two of the following characteristics: (1) homonymous visual symptoms and/or unilateral sensory symptoms C (2) over at least 5 minutes, at least one aura symptom develops gradually and/or different aura symptoms occur in succession (3) each aura symptom lasts between 5 and 60 minutes
Headache fulfilling criteria for migraine without aura (Table 1) begins either: D (1) during the aura, or (2) follows aura within 60 minutes
E At least 2 attacks fulfilling the above criteria
F Not attributed to any other disorder.
3 Of particular interest to this thesis is the involvement of the visual system in migraine. Many sufferers report visual triggers for their events (Hay et al. 1994; Mulleners et al. 2001a; Kowacs et al. 2004) and experience mild impairment of sight during an attack, such as blurring of vision (Hupp et al. 1989; Vincent et al. 1989). Furthermore, visual aura is the dominant type of aura, occurring in over 90% of MA attacks (Russell and Olesen 1996; Kelman 2004b). The classic visual aura – the scintillating scotoma, first described by Airy (1870) – is homonymous and hemianopic (Lashley 1941; Alvarez 1960; Hupp et al. 1989; Grüsser 1995; Kelman 2004b; Schott 2007), an example of which is pictured in Figure 1.1. The predominance of visual symptoms during attacks suggests that the pathophysiology of migraine, at least in part, involves the visual system.
1.2.2 Pathophysiology of the migraine attack
Currently, the pathophysiological processes responsible for the initiation of migraine are incompletely understood. Nevertheless, this section describes the theories that are generally accepted at present, which account for specific features of the migraine attack.
1.2.2.1 Brainstem activation
It is recognised that migraine attacks begin in the central nervous system, with evidence from neuroimaging techniques, including blood oxygen level dependent functional magnetic resonance imaging (fMRI-BOLD) and positron emission tomography (PET) (for a review, see Schwedt and Dodick 2009). These techniques are based on the notion of neurovascular coupling, whereby the vascular response adapts according to the metabolic demand of neurones (Jueptner and Weiller 1995; Heeger and Ress 2002).
4
Figure 1.1 – Progression of visual disturbance in migraine aura. Reprinted from Journal of the American Optometric Association, Volume 60, Bowerman, L. “Transient visual field loss secondary to migraine”. pp 912 – 916. Copyright (1989) American Optometric Association, with permission from Elsevier.
5 An increase in the intensity of the fMRI-BOLD or PET signal implies increased blood flow, which is generally interpreted as increased neural activity. This can occur due to primary neural activation or increased feedback (Jueptner and Weiller 1995; Heeger and Ress 2002). Studies that have captured images at the onset of a migraine are few, due to the generally unpredictable nature of attacks. Nevertheless, these studies have provided evidence for increased signal intensity at the brainstem before the presence of migraine symptoms using PET (Weiller et al. 1995; Bahra et al. 2001; Matharu et al. 2004; Afridi et al. 2005) and fMRI (Cao et al. 2002). In addition, Cao et al. (2002) showed that the brainstem signals preceded the appearance of increased signal intensity in the cortex. Initial activation of the brainstem is consistent with the role of the brainstem in the modulation of pain and nausea (Akerman et al. 2011), which are hallmark symptoms of the migraine attack, and in controlling motor functions such as yawning or gastrointestinal disturbances that can form part of the prodromal phase (Blau 1980; Kelman 2004b; Schoonman et al. 2006).
1.2.2.2 The migraine headache
Traditionally, a primary vascular mechanism was believed to underlie the migraine headache (Wolff 1963), with distension of the cerebral blood vessels causing the throbbing nature of the headache. However, vasodilation of the meningeal and/or extracranial arteries does not always provoke migraine headache (Kruuse et al. 2003; Rahmann et al. 2008; Schoonman et al. 2008), and the modest 9 to 12% dilation of cerebral blood vessels found is likely insufficient to activate pain receptors (Bolay et al. 2002).
The pain of migraine is now attributed to the activation and sensitisation of the trigeminovascular system (for reviews, see Moskowitz and Macfarlane 1993; Pietrobon 2005; Dalkara et al. 2006; Vecchia and Pietrobon 2012; Noseda and Burstein 2013). The trigeminovascular system has two major roles. It is responsible for the neural transmission of pain, and acts directly on blood vessels to maintain blood flow in the brain. Trigeminal nerve endings are densely packed throughout the
6 meninges. Activation of these nerve endings via the sensory afferent fibres of the trigeminal nerve results in release of vasoactive peptides (primarily calcitonin gene- related peptide), which cause vasodilation of the cerebral blood vessels, as well as other pro-inflammatory substances that lead to perivascular inflammation. Evidence for trigeminovascular involvement in the pain phase comes partly from the fact that the triptan class of drugs provides effective acute anti-migraine therapy (Monteith and Goadsby 2011). Triptans are agonists of serotonin (5-HT) receptors. Serotonin directly modulates the trigeminovascular system by causing vasoconstriction and subsequent inhibition of the release of pro-inflammatory neuropeptides. For example, it has been shown that increased levels of calcitonin gene-related peptide during migraine attacks (Ho et al. 2010) are returned to normal levels after acute treatment with sumatriptan, coinciding with subjective relief from pain (Goadsby and Edvinsson 1993).
In addition to the vascular response, activation of the trigeminal afferents sends pain signals through the brain stem, the thalamus, and to cortical areas involved in the perception and processing of pain. Serotonin is also implicated in the neurotransmission of these pain signals. Thus, it is not necessary that acute anti- migraine compounds terminate pain by acting on the vascular response. Rather, inhibition of signal transmission to the pain centres by way of 5-HT receptor agonists (Shepheard et al. 1999; Shields and Goadsby 2006) is also believed to play an important role in the acute treatment of migraine headache.
Not only is the neural network responsible for the processing of pain acutely involved during a migraine attack, but it is also believed to be dysfunctional in the interictal period. The increase in brainstem activity observed at the onset of migraine (Weiller et al. 1995; Bahra et al. 2001; Cao et al. 2002; Matharu et al. 2004; Afridi et al. 2005) is not solely due to the presence of pain during an attack. Rather, the brainstem signals remain increased even under effective analgesia (Weiller et al. 1995; Bahra et al. 2001; Afridi et al. 2005) and are persistently increased in people with chronic migraine (Matharu et al. 2004). Some investigators have therefore proposed that migraine headache also arises from dysfunction of the brainstem areas involved
7 in central control of pain, whereby normal sensory input is perceived as pain, known as ‘central sensitisation’ (reviewed by Akerman et al. 2011).
1.2.2.3 Cortical spreading depression
Cortical spreading depression (CSD) is believed to be the physiological substrate for migraine visual aura (Schwedt and Dodick 2009; Vecchia and Pietrobon 2012; Pietrobon and Moskowitz 2013). First described by Leão (1944), CSD is a slow, self- propagating wave of neuronal silence, which has been triggered by direct focal stimulation of the cerebral cortex in rabbit (Leão 1944) and rat (Lauritzen et al. 1982).
In humans, CSD has been inferred from neuroimaging of transient, regional reductions in blood flow that spread from the occipital cortex. Olesen et al. (1981) and Lauritzen et al. (1983) used intra-arterial Xenon-133 injections to demonstrate a focal area of reduced blood flow starting at the occipital pole, i.e. corresponding to the visual cortex. The reduction in blood flow gradually spread through the occipital cortex and coincided with the aura phase of migraine. The 20 to 25% reduction in blood flow did not exceed ischaemic levels and was therefore termed ‘oligaemia’. Likewise, Hadjikhani et al. (2001) imaged three people within 20 minutes of the beginning of a migraine visual aura. fMRI-BOLD was used to achieve higher spatial resolution than previously (Olesen et al. 1981; Lauritzen et al. 1983). On the contralateral hemisphere to the visual aura, the wave of oligaemia spread across the visual cortex at the same rate (approximately 3 mm/min, see also Lashley 1941; Lauritzen et al. 1983; Wilkinson 2004) and corresponded in size to the progression of the visual aura (see also Grüsser 1995; Schott 2007). The results did not provide conclusive evidence, however, that CSD begins in primary visual cortex (V1), with one person showing initiation of CSD in extrastriate visual area V3A (Hadjikhani et al. 2001).
8 The theory of spreading depression was initially assumed to account exclusively for migraine aura (Lauritzen 1994). However, there has been some evidence that CSD- like changes might also occur at the onset of MO attacks. Woods et al. (1994) made a serendipitous PET recording of regional cerebral blood flow, consistent with spreading depression in the visual cortex, minutes after the onset of a migraine headache in a patient undertaking scans for other reasons. As this person had never experienced an aura before, and on this occasion only reported a temporary difficulty in focusing, this could not be diagnosed as a typical visual aura (International Headache Society 2013). Later studies have provided conflicting evidence, neither proving nor disproving the presence of CSD in migraine without aura. In support of the observation by Woods et al. (1994), Denuelle and colleagues also used PET to demonstrate reductions in blood flow in seven individuals with MO (Denuelle et al. 2008). This raises the possibility of a ‘silent aura’, whereby CSD might be confined to subcortical regions (such as the hippocampus, see Kunkler and Kraig 2003) and therefore does not manifest as aura. However, other studies during attacks of MO have not detected CSD-like changes in cerebral perfusion (Ferrari et al. 1995; Sanchez del Rio et al. 1999). Thus, it is unclear whether CSD is involved in the pathogenesis of both MO and MA attacks. Apart from the presence of aura, the two subtypes generally share the same clinical characteristics such as trigger and precipitating factors, nausea, headache, and pro- and post-dromal symptoms (International Headache Society 2013). The headache appears to be identical in symptomatology, although the pain phase of migraine without aura is often more protracted and severe (Rasmussen and Olesen 1992). An added complexity is that some individuals will experience both major forms of migraine at different times, with estimates of co-existing MO and MA ranging from 8% (Rasmussen and Olesen 1992) to 29% (Cologno et al. 1998). Whether MO and MA are separate conditions remains a point of controversy (Manzoni and Torelli 2008; Purdy 2008). As such, there is no compelling evidence for or against pooling of migraine subtypes. Nevertheless, in this thesis, attempts have been made to discuss the findings in the literature as distinct between MO and MA groups, as these are separate diagnoses in the classifications of the International Headache Society (2013).
9 A further controversy that has not been resolved to date is whether CSD leads to the triggering of migraine headache and other associated symptoms. Experiments in rodents show that CSD can activate the trigeminovascular system and induce the release of pro-inflammatory and vasoactive substances that lead to pain (Bolay et al. 2002). However, a causal link between CSD and migraine headache is not supported by clinical observations (Wolthausen et al. 2009). Some people experience migraine aura without a headache (International Headache Society 2013) and aura does not always precede the headache, but can occur during or even after the headache in up to 3% of migraine sufferers (Russell and Olesen 1996). Thus, CSD does not currently provide a single, unifying concept that explains why people are susceptible to the generation of migraine symptoms.
1.2.3 Genetics of migraine
At present, the genetics and inheritance of migraine are poorly understood. The heritability of migraine is estimated to be, at best, 50% (Maher and Griffiths 2011). However, a rare form of migraine – familial hemiplegic migraine (FHM) – is known to be an autosomal dominant trait. FHM is characterised by motor aura symptoms, i.e. motor weakness or paralysis (International Headache Society 2013). Besides the motor aura and the possibility for more severe symptoms such as diffuse encephalopathy, loss of consciousness, seizures, and prolonged headache (Ducros et al. 2001), typical FHM attacks resemble attacks of MA (International Headache Society 2013). Moreover, attacks of FHM and MA can occur in the same people. Hence, studies of FHM in humans and animal models have provided some clues as to possible pathogenetic mechanisms of migraine, at a neurobiological level.
Animal studies using FHM knock-in mice support the view of a critical role of CSD in the initiation of the migraine attack. Three variants of familial hemiplegic migraine (FHM1, FHM2, and FHM3) are known to be due to mutations in the CACNA1A, ATP1A2, and SCNA1A genes, respectively. It is beyond the scope of this literature review to discuss in detail the molecular differences between these variants of FHM
10 (for more recent reviews, see Pietrobon and Moskowitz 2013 and Vecchia and Pietrobon 2012). Nevertheless, a common denominator of these FHM animal models is that the mutations result in excessive glutamate levels and altered glutamatergic transmission in the cortex, leading to impaired recovery from neural excitation and therefore recurrent excitation (Vecchia and Pietrobon 2012; Pietrobon and Moskowitz 2013). This is believed to create an abnormal resting state of ‘hyperexcitability’ and the physiological conditions for ignition of CSD in FHM, which is potentially relevant to the triggering of migraine in general (Vecchia and Pietrobon 2012; Pietrobon and Moskowitz 2013).
1.3 Cortical hyperexcitability
As migraine is considered a disorder of the central nervous system (Schwedt and Dodick 2009; Vecchia and Pietrobon 2012; Pietrobon and Moskowitz 2013), investigations have considered the function of the brain in order to gain insight into its pathophysiology. Vision anomalies are the focus of this thesis; hence, the remainder of this literature review is devoted to a detailed discussion of studies that have tested the visual system of migraine sufferers during the interictal period. Most of the findings are in keeping with the genetic studies of familial hemiplegic migraine (Vecchia and Pietrobon 2012; Pietrobon and Moskowitz 2013) that imply a mechanism of cortical ‘hyperexcitability’ (Welch et al. 1990; Aurora and Wilkinson 2007; Coppola et al. 2007c) in migraine, as discussed below.
1.3.1 Evidence for hyperexcitability from the visual cortex
1.3.1.1 Transcranial magnetic stimulation
The propensity of cortical neurones to respond to direct stimulation has been measured by transcranial magnetic stimulation (TMS). After the application of brief magnetic pulses, neurones under the site of stimulation depolarise and discharge
11 action potentials. When TMS is applied over the visual cortex, phosphenes (flashes of light are elicited (for a review, see Walsh and Cowey 1998). In studies of people with migraine, the absolute threshold for inducing the perception of such phosphenes has been used to infer the level of neuronal activity at the visual cortex (Aurora et al. 1998; Battelli et al. 2002; Bohotin et al. 2003; Siniatchkin et al. 2009).
The majority of TMS studies of the visual cortex in migraine have delivered electric currents over the occipital lobe, which is presumed to elicit neural spiking activity predominantly in V1. A detailed discussion of individual TMS reports is beyond the scope of this review. Nevertheless, across multiple studies and stimulation paradigms, people with migraine and visual aura consistently demonstrate greater susceptibility to the induction of phosphenes, suggesting hyperexcitability of area V1 (for a recent meta-analysis, see Brigo et al. 2013). Evidence for similar effects in people who do not suffer from aura, however, is equivocal (Brigo et al. 2013).
TMS can be delivered to more specific brain areas using fMRI scans to localise the region of interest. Only one study to date has delivered TMS over extrastriate visual area V5/MT, which is located in the ventrolateral occipital cortex (Dumoulin et al. 2000). In that study, phosphenes were elicited at a lower threshold in both aura and non-aura groups (Battelli et al. 2002). In addition, TMS has also demonstrated hyperexcitability of extrastriate visual area V3A (Fierro et al. 2003). These findings argue for a generalised cortical abnormality in migraine, involving both primary and extrastriate visual areas.
The studies described above using TMS of the visual cortex have necessarily involved the subjective reporting of phosphenes. To supplement these findings, functional imaging techniques that are considered to be more objective physiological measures have strengthened the concept of hyperexcitability in migraine, as discussed next.
12 1.3.1.2 Brain neuroimaging
Brain neuroimaging techniques such as fMRI have been used to image the visual cortical changes that occur at the onset of, and during an attack, as detailed earlier in this chapter and reviewed by Schwedt and Dodick (2009). More common, however, has been the use of brain neuroimaging in the interictal period to explore cortical function in people with migraine. Presuming a hyperxcitable state, investigators have predicted increased signal intensities in the visual cortex of migraine sufferers in response to visual stimulation.
A number of studies have asked participants to describe their perception of high- contrast grating patterns, demonstrating that people with migraine report more discomfort and illusions of colour, shape, and motion than controls (Khalil 1991; Coleston and Kennard 1993; Oelkers et al. 1999; Shepherd 2000). The subjective nature of these reports, however, raises a potential bias, as migraine sufferers have a greater tendency for hypochondria (Burker et al. 1989) and may therefore have a preconceived judgment of the level of discomfort experienced. To supplement these subjective measures of hypersensitivity, Huang et al. (2003) conducted fMRI-BOLD in response to high-contrast (75%), square-wave grating patterns. Six MA sufferers and six age-matched control participants were presented with gratings of different spatial frequency, ranging from 0.25 to 19.2 cycles/degree. Consistent with previous reports (Khalil 1991; Coleston and Kennard 1993; Oelkers et al. 1999; Shepherd 2000), the people with migraine reported more illusions than controls for gratings of 1 to 4 cycles/degree (Huang et al. 2003). Additionally, these same patterns provoked the greatest increase in fMRI-BOLD signal intensity at the visual cortex, implying an association between physiological and behavioural responses, both of which are indicative of hyperexcitability in migraine.
Because the typical scintillating scotoma of visual aura (Figure 1.1) has often been described as expanding zigzag lines (for a review, see Schott 2007), it is possible that neurones of the visual cortex involved in the processing of orientation information
13 are more likely to exhibit hyperexcitable behaviour in people with visual aura. Orientation tuning has been well documented in cortical neurones (Hubel and Wiesel 1968), with a high density of neurones that are selectively stimulated by oriented patterns in V1 and V2 (Tootell et al. 1998). Thus, Vincent et al. (2003) hypothesised that orientation-selective neurones in V1 and V2 of MA sufferers would be abnormally responsive to tilted white bars on a black background. Unlike the five control participants who largely showed activation of the orbitofrontal cortex, the five MA participants demonstrated enhanced fMRI-BOLD signals in Brodmann areas 17 and 18, corresponding to areas V1 and V2, with one person showing hyperactivity of the extrastriate visual cortex in Brodmann area 19 (Vincent et al. 2003).
A later fMRI study by Antal et al. (2011) confirmed that hyperexcitability of the visual cortex in migraine is not confined to V1 and V2. In that study, a random dot motion stimulus was used to explore global motion perception (Watamaniuk et al. 1989). The stimulus consisted of white moving dots on a black background, presented over several frames. A percentage of dots moved coherently in a dominant direction (signal dots), while the remaining dots (noise dots) moved in random directions. The dots that showed the dominant direction of motion were randomly chosen in each frame of the movie sequence. Thus, individual dots could not be tracked from frame to frame, which removed any local motion cues in determining the overall coherent motion direction. In response to the random dot stimulus, the people with migraine (12 MO, 12MA) demonstrated higher intensity of fMRI-BOLD signals at extrastriate visual area V5/MT (Antal et al. 2011), which is the visual area that is considered necessary for the global perception of motion. Support for this view comes from single-cell neurophysiological studies in primates (Newsome et al. 1989; Pack et al. 2001), human fMRI studies that have identified activation of area V5/MT during the perception of motion (Greenlee 2000; Paradis et al. 2000), and psychophysical studies that have demonstrated impaired motion perception following lesions to area V5/MT in primates and humans (Newsome and Pare 1988; Baker et al. 1991; Schiller 1993). Note that thus far in this discussion, no distinction has been made between MO and MA groups in studies that have investigated area
14 V5/MT using fMRI and global dot motion stimuli (Antal et al. 2011) or by direct stimulation using TMS (Battelli et al. 2002).
Neuroimaging techniques have also been used to measure relative concentrations of brain metabolites, where an increase in markers of brain metabolic activity has been considered as a corollary of hyperexcitability in migraine. Sandor et al. (2005) imaged MA patients using magnetic resonance spectroscopy (MRS) before, during, and after visual stimulation with a checkerboard pattern. The MRS image was divided into a matrix comprising 64 voxels. Only voxels that were classified as containing ‘visual’ or ‘non-visual’ cortical areas were compared. For the visual voxels, MRS found higher resting lactate levels in the five people with MA who suffered exclusively from attacks with visual aura. In addition, during checkerboard presentations, lactate levels remained consistently elevated. A further five MA participants with additional non- visual symptoms (numbness, tingling, dysphasia) as part of their migraine aura were imaged, showing significantly elevated levels of lactate in response to visual stimulation, but not in the resting period (Sandor et al. 2005). Thus, the authors interpreted these results as indicating hyperexcitability of the visual cortex in MA sufferers, whose presumed CSD events explicitly involve the visual cortex. Consistent with these findings is a tendency for increased lactate levels in response to photic stimulation in people with aura, but not in MO sufferers, as reported by Sarchielli et al. (2005).
A limitation of these brain neuroimaging studies is the relatively small sample sizes, with some studies only testing people with aura (Huang et al. 2003; Vincent et al. 2003; Sandor et al. 2005). Furthermore, the increase in blood flow measured by fMRI-BOLD or the increase in brain metabolites measured by MRS does not necessarily imply an increase in spiking activity of cortical neurones. To assess both the neuronal and vascular response in migraine, Zaletel et al. (2005) simultaneously recorded cerebral arterial blood flow velocity by Doppler and visual evoked potentials, otherwise known as the visual evoked response (VER), to a checkerboard stimulus. The VER is an electrophysiological technique that measures the electrical potentials generated by the synchronised spiking activity of a population of cortical
15 neurones in response to visual stimulation. Consistent with previous work in humans using fMRI (Arthurs et al. 2000) and near infrared spectroscopy (Obrig et al. 2002), Zaletel et al. (2005) showed an increase in inferred neuronal activity and the haemodynamic response in control and migraine groups (16 MA, 14 MO). However, based on linear regression analysis, the migraine group showed a larger change in blood flow per unit change in the VER, relative to controls (Figure 1.2). The implication of the study by Zaletel et al. (2005) is that for a given level of neural activity, the vascular response in people with migraine is disproportionate to what is expected in normal observers. Studies that have inferred the level of neural activity in migraine using the VER, without measurement of the vascular response, are described next.
Figure 1.2 – Altered neurovascular coupling in migraine. The visually evoked cerebral blood flow velocity (cm/s) is plotted as a function of VER amplitude (µV) in patients with migraine (n = 30, unfilled circles) and healthy participants (n = 30, filled circles). The dashed line represents the linear regression line for the migraine group data (r = 0.63, slope = 0.88), and the solid line represents the linear regression lines for the control group data (r = 0.66, slope = 0.55). Reprinted from Cephalalgia, Volume 25, Zaletel, M., Strucl, M., Bajrovic, F.F., and Pogacnik, T. “Coupling between visual evoked cerebral blood flow velocity responses and visual evoked potentials in migraneurs.” pp 567 – 574. Copyright (2005), with permission from Sage Publications.
16 1.3.1.3 Visual evoked potentials (visual evoked response)
Visual evoked potentials, or the visual evoked response (VER), are understood to be the summation of cortical excitatory and inhibitory post-synaptic potentials (Eccles 1951), arising from cortical neurones whose apical dendrites are perpendicularly oriented relative to the scalp surface (Fahle and Bach 2006). Specifically, the VER is detected by placing electrodes on the scalp overlying the visual cortex on the midline, but also frequently by placing electrodes lateral to the midline (Odom et al. 2010). Using fMRI, the earliest neural generators of the VER in terms of timing relative to stimulus onset have been localised to V1, with subsequent activation of the extrastriate visual areas (Di Russo et al. 2005; Di Russo et al. 2007).
The VER waveform is characterised by its magnitude (amplitude) and timing (Odom et al. 2010). Hyperexcitability would predict a larger amplitude or faster timing of the VER. In support of this hypothesis, earlier studies of people with migraine used photic stimulation (flash VER), most of which demonstrated increased amplitude in response to flashes and flickering light (Lehtonen 1974; MacLean et al. 1975; Connolly et al. 1982; Nyrke and Lang 1982; Brinciotti et al. 1986; Nyrke et al. 1989; Khalil 1991).
The majority of VER studies in migraine to date have measured the pattern-reversal VER (PVER), which uses an achromatic patterned stimulus with alternating light and dark regions (e.g. a checkerboard) to invoke activity of contrast-sensitive neurones. As reviewed by Ambrosini et al. (2003) and Magis et al. (2007a), studies of the PVER in migraine have almost exclusively used stimuli of high contrast above 80% in order to generate a robust response. Consistent with cortical hyperexcitability, several investigators have found abnormally increased amplitude (Kennard et al. 1978; Diener et al. 1989; Mariani et al. 1990; Shibata et al. 1997; Shibata et al. 1998; Lahat et al. 1999; Khalil et al. 2000; Marrelli et al. 2001; Shibata et al. 2008) or faster PVER responses (Aloisi et al. 1997; Marrelli et al. 2001; Spreafico et al. 2004) in migraine. Furthermore, as illustrated in Figure 1.3, studies have shown deficient habituation in
17 migraine sufferers, whereby the amplitude of the PVER is augmented and/or stays elevated over time with continuous stimulation for at least two minutes (Schoenen et al. 1995; Afra et al. 1998; Wang et al. 1999; Bohotin et al. 2002; Ozkul and Bozlar 2002; Di Clemente et al. 2005; Fumal et al. 2006; Coppola et al. 2007b; Magis et al. 2007b; Coppola et al. 2010a; Coppola et al. 2010b; Coppola et al. 2011; Shibata et al. 2011; Coppola et al. 2013; Vigano et al. 2013). In contrast, the PVER amplitude declines with repeated stimulation in normal observers (Janz et al. 2001; Omland et al. 2011).
Figure 1.3 – Deficient habituation of the visual evoked response in migraine. Example waveforms depicting (A) Normal habituation in a healthy volunteer (B) Deficient habituation in a migraine sufferer. In studies of habituation, responses are recorded continuously, typically for at least two minutes. The top waveform in each panel represents the first response recorded, whereas the bottom waveform is the last response recorded. The height of the shaded area indicates the amplitude of the response. Reprinted from Neurobiology of Learning and Memory, Volume 92, Coppola, G., Pierelli, F., and Schoenen, J. “Habituation and migraine.” pp 249 – 259. Copyright (2009) with permission from Elsevier.
As migraine symptoms are often lateralised to one side, investigators have also considered the possibility for asymmetry in cortical hyperexcitability. Thus, the PVER has been recorded with electrodes placed over the right and left hemispheres of the occipital lobe. Using this method, a number of studies have reported greater interhemispheric asymmetry of the PVER in migraine (Tsounis et al. 1993; Tagliati et
18 al. 1995; Shibata et al. 1997; Shibata et al. 1998; Logi et al. 2001; Coppola et al. 2007a). The abnormality is presumed to coincide with the side of greater hyperexcitability, such as higher amplitudes on the side contralateral to the reported visual aura (Tagliati et al. 1995; Shibata et al. 1997; Shibata et al. 1998), or increased PVER amplitude (Logi et al. 2001) and decreased peak times (Logi et al. 2001) on the same side as the headache. However, some studies have failed to find an association between greater asymmetry and specific migraine characteristics (Mariani et al. 1990; Khalil 1991; Coppola et al. 2007a), because of the variable nature of migraines. The migraine aura or headache is not confined to the same side for every migraine attack (Coppola et al. 2007a), and approximately one-third of people with visual aura report bi-hemifield involvement (Russell and Olesen 1996).
Electrophysiological abnormalities in migraine appear to be temporally tuned. Habituation defects in migraine have been demonstrated with a reversal rate of at least 3 Hz (Schoenen et al. 1995; Afra et al. 1998; Wang et al. 1999; Bohotin et al. 2002; Ozkul and Bozlar 2002; Di Clemente et al. 2005; Fumal et al. 2006; Coppola et al. 2007b; Magis et al. 2007b; Coppola et al. 2010a; Coppola et al. 2010b; Coppola et al. 2011; Shibata et al. 2011; Coppola et al. 2013; Vigano et al. 2013), but not under conditions of lower temporal frequency (Oelkers et al. 1999; Sand and Vingen 2000; Oelkers-Ax et al. 2005). Similarly, studies have measured the flash VER across a range of temporal frequencies, demonstrating a trend for more pronounced abnormalities under conditions of higher temporal frequency of 18 to 20 Hz (Nyrke and Lang 1982; Nyrke et al. 1989; Khalil 1991). The strength of these latter studies is that the flash VER was measured in the same group of participants.
Depending on the temporal frequency of pattern reversal, the PVER can take on two characteristic waveforms. When elicited with a pattern alternating at temporal frequencies below approximately 3 Hz (6 reversals/second), a ‘transient’ response is recorded (for example, see Figure 1.3) (Odom et al. 2010). At higher temporal frequencies of stimulation (≥ 4 Hz, or 8 reversals/second), the successive waveforms begin to overlap and a sinusoidal-like ‘steady-state’ response is generated (Odom et al. 2010). Note that the terms ‘sustained’ and ‘transient’ are often used to mean the
19 opposite, outside of electrophysiology. Steady-state responses are presumed to measure similar mechanisms as behavioural measures of temporal processing (King- Smith and Kulikowski 1975). Given the possibility for temporal tuning of VER abnormalities in migraine, it might be that more pronounced deficits are found in the steady-state PVER. Evidence to this effect, however, is equivocal, given two previous studies that have recorded transient and steady-state responses in the same cohort of migraine patients. Diener et al. (1989) recorded responses to 1.56 Hz (transient) and 8.33 Hz (steady-state) and reported significantly increased amplitudes under both conditions in a large cohort of migraine sufferers (54 MO, 4 MA). In contrast, Marrelli et al. (2001) reported a significant increase in steady-state PVER amplitude in their cohort of 34 children with migraine, yet the increase in the amplitude of the transient response did not reach statistical significance.
Another point of controversy in the VER literature is the heterogeneity of findings (Ambrosini et al. 2003; Magis et al. 2007a). Unlike the TMS and neuroimaging results noted earlier, the electrophysiological evidence in support of a cortical hyperexcitability in migraine is less consistent. Studies have yielded conflicting results, sometimes in the same cohort of migraine sufferers. One of the earliest studies by Kennard et al. (1978) recorded the PVER in 28 patients who, on average, had suffered migraines with visual aura for 18 years. The two major findings were increased amplitude in a subset of patients with left-sided headache, yet prolonged timing of the PVER across all patients. One of the reasons put forth to explain the delay is that repeated episodes of visual aura might accelerate the natural progression of cell loss and degeneration (Kennard et al. 1978). This explanation was also adopted by Khalil et al. (2000), who found increased PVER amplitudes in people with a short history of MA (ten years or less), but decreased amplitudes in people who had suffered attacks with aura for at least 30 years. Delayed responses, also found by several other investigators (Diener et al. 1989; Khalil and Legg 1989; Mariani et al. 1990; Sener et al. 1997; Oelkers et al. 1999), as well as reduced amplitudes (Polich et al. 1986; Khalil et al. 2000; Boylu et al. 2010), are not suggestive of a cortical hyperexcitability.
20 It is difficult to reconcile findings of previous studies in migraine because there are a number of factors that contribute to variability in the VER. Alterations in the visual evoked response can be due to anatomical changes in the brain or earlier in the visual pathway, as the VER depends on the integrity of the retinogeniculostriate pathway (Fahle and Bach 2006). The timing and amplitude of the waveform is influenced by age (Dustman and Beck 1969; Celesia et al. 1987) and stimulus characteristics such as contrast (Spekreijse et al. 1973) and spatial and temporal frequency (Heine and Meigen 2004). Furthermore, anatomical differences in the position of the visual cortex relative to the recording electrode can contribute to inter-individual variability (Brindley 1972; Steinmetz et al. 1989; Hood and Zhang 2000; Ishikawa et al. 2011). Finally, although the VER is considered to be an objective measure; it is also significantly affected by human factors such as loss of attention and voluntary defocus (Douthwaite and Connor 1989; Mezer et al. 2004). Thus, a major purpose of this thesis is to investigate possible explanations for altered VER in people with migraine, as outlined in the overview of the thesis at the end of this chapter.
1.3.1.4 Hypersensitivity to visual stimuli
The TMS, brain neuroimaging, and electrophysiological methods described above purport to measure responses from the visual cortex using equipment placed over the occipital lobe. A possible behavioural corollary of visual cortical hyperexcitability, as implied by the fMRI study by Huang et al. (2003), is hypersensitivity to visual stimuli. In between attacks, people with migraine often report heightened awareness to striped patterns and flickering lights (Hay et al. 1994; Kowacs et al. 2004) and show more aversive behaviour when viewing high-contrast grating patterns (Wilkins et al. 1984; Marcus and Soso 1989; Khalil 1991; Coleston and Kennard 1993; Shepherd 2000; Huang et al. 2003). Similarly, when participants have been asked to increase the contrast of a flickering computer screen until they can no longer look at it comfortably, people with migraine abort the test at a lower contrast level than non- headache controls (Wilkinson et al. 2008; Karanovic et al. 2011; Thabet et al. 2013). Hypersensitivity to visual stimuli is particularly evident in people with migraine at the
21 peak of the contrast sensitivity function – 3 to 4 cycles/degree spatial frequency (Wilkins et al. 1984) and 10 Hz temporal frequency (Wilkinson et al. 2008).
These observations suggest an abnormal increase in neural gain in migraine. Neural gain refers to the response behaviour of sensory systems as a function of stimulus intensity. Neural gain is a network mechanism, whereby the response is ‘divided’ or scaled due to inhibition from the surrounding neurones (Heeger 1992; Carandini and Heeger 1994; Wilson et al. 2012). For example, contrast gain is the neuronal response as a function of stimulus contrast. At a single-cell level, normal contrast gain shows a non-linear characteristic in primate visual cortex (Albrecht and Hamilton 1982; Sclar et al. 1989). The neurone’s response increases with stimulus intensity, until saturation at high contrast, i.e. maximum firing rate. This has also been shown in human VER measurements in normal observers (see Figure 1.4, filled circles and also Spekreijse et al. 1973; Ross and Speed 1991), representing the massed response across a large population of cortical neurones.
Figure 1.4 – Altered contrast gain of the visual evoked response. Group mean PVER amplitude plotted as a function of contrast in patients with epilepsy (unfilled circles) and controls (filled circles). Error bars represent the standard error of the mean. Reprinted from Annals of Neurology, Volume 70, Tsai, J.J., Norcia, A.M., Ales, J.M., Wade, A.R. “Contrast gain control abnormalities in idiopathic generalized epilepsy.” pp 574 – 582. Copyright (2011) American Neurological Association, with permission from John Wiley and Sons.
22 In people with epilepsy, PVER contrast gain is increased (Porciatti et al. 2000; Tsai et al. 2011). That is, the amplitude of the PVER does not saturate, but continues to increase with contrast (Figure 1.4, unfilled circles). Heightened sensitivity to visual stimuli in people with migraine might similarly result from increased gain, such that a normally innocuous visual stimulus elicits a much larger response. It remains to be determined whether neural gain is altered in people with migraine.
1.3.1.5 Psychophysical thresholds
A question that has not been addressed by the brain neuroimaging and electrophysiological techniques described earlier, which rely on visual stimulation that is suprathreshold (e.g. high contrast), is whether people with migraine show increased responses to very weak, or even non-existent stimuli. Thus, psychophysical tests of vision have measured responses at threshold, using tests that are believed to require processing of visual properties at a cortical level.
The orientation discrimination threshold is the smallest angular difference in orientation that can be perceived. Performance on this task requires processing by neurones that are tuned to orientation. Such cells are found in the visual cortex (Hubel and Wiesel 1968), whose orientation-selectivity is believed to arise, in part, from a significant feedforward contribution from the lateral geniculate nucleus (LGN) of the thalamus (Vidyasagar and Urbas 1982). Elevated orientation discrimination thresholds have been found in both aura and non-aura groups (McKendrick et al. 2001; Tibber et al. 2006), although the deficits in MA participants did not reach statistical significance in the study by Wilkinson and Crotogino (2000).
A number of studies in migraine have measured contrast sensitivity (i.e. the minimum detectable difference between light and dark features) using grating patterns of different spatiotemporal frequencies, demonstrating elevated contrast detection thresholds in both MO and MA groups when measured foveally (Khalil 1991; Benedek et al. 2002; Yenice et al. 2007; Shepherd et al. 2012). Contrast
23 sensitivity deficits have also been found across the visual field (typically across the central 20° to 30°) in a significant proportion of people with migraine (approximately 35 to 50%) (Lewis et al. 1989; Drummond and Anderson 1992; De Natale et al. 1993; Nizankowska et al. 1997; McKendrick et al. 2000; McKendrick et al. 2002; Çomoglu et al. 2003; McKendrick and Badcock 2004a; McKendrick and Badcock 2004b; McKendrick and Badcock 2004c; Yenice et al. 2006). However, contrast sensitivity losses in migraine are relatively non-specific regarding the anatomical locus for dysfunction. Contrast sensitivity requires processing at several levels of the visual pathway, as lesions of the retina, LGN, or cortical area V1 have shown profound contrast sensitivity losses in humans (Koerner and Teuber 1973) and primates (Merigan and Eskin 1986; Merigan et al. 1993).
Unlike the contrast sensitivity measures noted above, which are not critically affected by lesions to visual area V2 (Merigan et al. 1993), global form and global motion stimuli are believed to assess extrastriate cortical visual processing. Physiological insights into the neural substrates of global shape perception have come from studies in primates (Kobatake and Tanaka 1994; Gallant et al. 1996), with parallel evidence for involvement of extrastriate area V4 in humans (Wilson and Wilkinson 1998; Gallant et al. 2000; Wilkinson et al. 2000). Global form detection thresholds have been found to be elevated in people with migraine (Ditchfield et al. 2006; McKendrick et al. 2006b) using variants of the Glass pattern paradigm (Glass 1969). The stimuli used by Ditchfield et al. (2006) and McKendrick et al. (2006b) consisted of pairs of elements (e.g. dots, lines), a percentage of which (signal) were oriented 90° relative to a line drawn from the centre of the stimulus, creating the perception of a concentric shape. The remaining elements were oriented randomly (noise). Because all orientations were represented, the shape of the Glass patterns could not be detected using local orientation cues (Wilson and Wilkinson 1998). Similarly, Wagner et al. (2013) used radial frequency patterns to demonstrate impaired global form perception in people with migraine. These patterns are defined by sinusoidal modulation of the radius of a base circle in polar coordinates, creating small deviations from circularity depending on the amplitude of modulation (Wilkinson et
24 al. 1998). The visual system must sum all available orientation information to be able to identify the global shape (Wilkinson et al. 1998).
As described earlier in this chapter, global dot motion refers to the dominant direction of motion observed in a field of moving dots. The global dot motion stimulus removes local motion cues by randomising the dots that display the dominant motion direction in each frame (Watamaniuk et al. 1989). The ability to detect the global motion depends on the level of motion coherence, i.e. the percentage of dots moving in the dominant direction of motion. Studies have consistently demonstrated that migraine sufferers, with and without aura, require more motion coherence for the dominant direction of motion to be correctly identified, measured both foveally (McKendrick et al. 2001; McKendrick and Badcock 2004c; Antal et al. 2005; Ditchfield et al. 2006; McKendrick et al. 2006b; Shepherd et al. 2012) and peripherally (McColl and Wilkinson 2001; McKendrick et al. 2001; McKendrick and Badcock 2004c).
Taken together, this collection of studies demonstrates elevated thresholds in both aura and non-aura groups. On the basis of these psychophysical findings, there is little evidence for hypersensitivity of threshold responses in migraine.
1.3.2 Changes in cortical hyperexcitability
Thus far, this literature review has focused on cross-sectional reports during the interictal period. However, migraine is an episodic condition, with distinct interictal and ictal phases. Accordingly, studies have considered whether cortical visual responses might reflect changes in hyperexcitability in temporal relation to a migraine event, as described in this section.
Shibata et al. (1998) recorded the PVER at a single visit. The cohort of 39 migraine participants was divided into subgroups based on the number of days since the last migraine attack. PVER amplitude was increased relative to controls in the interictal
25 period. Furthermore, there was a trend for decreasing PVER amplitude with time post-migraine – amplitudes were highest one to 10 days after migraine, and lowest 21 to 30 days after migraine (Shibata et al. 1998). To explain this trend, the authors suggested that the people who were tested 21 to 30 days after a migraine were most likely those who were closer to an impending attack, implying that the PVER might normalise in the lead up to an attack. This possibility could not be directly tested, however, as data regarding the timing of the test visit relative to the next migraine were not collected.
An improvement to the study by Shibata et al. (1998) was the measurement of the PVER at multiple time points in the migraine cycle: before, during, and after (Judit et al. 2000; Sand et al. 2008; Sand et al. 2009). Judit et al. (2000) found deficient habituation in a group of people whose test visit was scheduled at least three days before and after a migraine. In contrast, a normal pattern of habituation was evident in a separate group of people tested the day before an attack, as well as another group tested during an attack (Judit et al. 2000). Similarly, PVER amplitudes were found to be abnormally elevated in the pre-attack period (within 72 hours of a migraine) and normalised with the onset of a migraine (Sand and Vingen 2000; Sand et al. 2008; Sand et al. 2009). An advantage of the latter studies by Sand and colleagues was that the PVER was measured at multiple time points in the same individuals (Sand et al. 2008; Sand et al. 2009). Such a study design provides more compelling evidence of visual changes in temporal relation to a migraine, by taking into account inter-individual variability.
Convergent evidence for normalisation of visual cortical activity in the lead up to a migraine has come from TMS and magnetoencephalography. Chen et al. (2009) recorded the magnetic fields produced by electric current at V1 in response to checkerboard stimuli. Normalisation of cortical responses was observed in the group that was tested in the peri-ictal period (within two days before or after a migraine), compared to the group tested later during the interictal period. Siniatchkin et al. (2009) studied the same individuals with migraine using TMS over the occipital lobe to elicit phosphenes. The migraine sufferers showed a significant increase in
26 phosphene thresholds 1 to 2 days before a migraine attack, relative to their thresholds measured in the interictal period. That is, the response became relatively more normal at migraine onset. Thus, taken together, it appears that the intensity of interictal cortical abnormalities, suggestive of hyperexcitability, reaches a maximum in the pre-attack period and subsequently normalises at the onset of migraine.
1.3.3 Investigations of mechanisms of hyperexcitability
The previous discussion highlights the evidence for cortical hyperexcitability as a key characteristic of migraine, by demonstrating that neural and vascular responses to visual stimulation are increased. However, what hyperexcitability means in terms of an exact neural basis is still not clear. This section describes the investigations that have provided some insight into possible dysfunctional processes that lead to cortical hyperexcitability in migraine.
Given there are changes in hyperexcitability over the migraine cycle (Judit et al. 2000; Sand et al. 2008; Chen et al. 2009; Sand et al. 2009; Siniatchkin et al. 2009), hyperexcitability is likely to involve dynamic mechanisms, such as changes in excitation and inhibition that are modulated by central neurotransmitter levels (Aurora and Wilkinson 2007; Coppola et al. 2007c). Key excitatory and inhibitory neurotransmitters that have been implicated in migraine are glutamate (Vikelis and Mitsikostas 2007), GABA (Puppe and Limmroth 2007), serotonin (Panconesi 2008), and dopamine (Barbanti et al. 2013). However, the evidence for altered levels of these neurotransmitters in migraine sufferers is equivocal. This is partly because direct measurement of brain neurotransmitter concentrations is not possible without invasive procedures. Instead, glutamate, serotonin, and dopamine levels have been inferred from varying sources, e.g. cerebrospinal fluid or peripheral blood and platelet samples (Vikelis and Mitsikostas 2007; Panconesi 2008; Barbanti et al. 2013), whereas levels of GABA have been inferred from clinical trials assessing efficacy of migraine prophylactic drugs (Puppe and Limmroth 2007).
27 An alternate way to indirectly study brain neuromodulatory status is to test psychophysical performance on visual tasks that are known to involve excitatory and inhibitory processes. Simplistically, if hyperexcitable neurones respond more strongly to suprathreshold stimuli, then they might also respond to otherwise sub-threshold stimuli. Alternatively, hyperexcitability might cause increased spontaneous neural firing and therefore increased noise. Hyperexcitability could also arise secondarily to decreased inhibition, or there might be some imbalance between inhibitory and excitatory processes. These possibilities have been explored in psychophysical experiments, as discussed below.
1.3.3.1 Increased noise
A possible explanation for reduced sensitivity on psychophysical tasks put forth by McKendrick and Badcock (2004c) is that cortical hyperexcitability might result in an increase in noise. For example, for the global dot motion stimulus (Watamaniuk et al. 1989), detection of global motion direction depends on the ability to distinguish those dots containing direction information (i.e. the signal) from randomly moving dots (i.e. the noise). Antal et al. (2005) showed that global motion perception was only impaired in the presence of incoherently moving dots, i.e. distractors. When the distractors were moving in the same direction as the signal dots, performance was not different between migraine and control groups. Hence, an increase in the proportion of distractors, or level of noise, could interfere with the perception of global motion in people with migraine.
Noise can be defined in terms of external or internal noise; the former being related to the stimulus characteristics, whereas the latter is defined as the proportion of the neural response that is unrelated to the presence of a signal (Barlow 1956). Internal noise is presumably produced by spontaneous neural activity, resulting in significant variability of neuronal responses. Random fluctuations in the neuronal response, however, may be independent of the presence of an ‘external’ noisy environment (‘additive’ noise) or require external noise before their effect becomes manifest
28 (‘multiplicative’ noise). More recently, perceptual studies in people with migraine have attempted to quantify the level of internal noise, and the effect of external noise on perceptual thresholds. Even when these studies have found higher levels of internal noise, whether the perceptual abnormality was dependent or independent of external noise (i.e. additive or multiplicative internal noise, respectively) varies. Wagner et al. (2010) measured contrast detection thresholds for a luminance target superimposed on a background of static noise (external noise), in which the luminance of each pixel was randomised to create a stippling effect. High levels of external noise (i.e. high contrast background) had a more detrimental effect in people with migraine (Wagner et al. 2010). On the other hand, similar thresholds were observed in the absence, or at low levels, of external noise. These results indicated that impaired contrast sensitivity was related to increased multiplicative internal noise, and were later replicated by Webster et al. (2012).
However, Webster and colleagues did not find increased multiplicative noise accounting for impaired perception on a range of global perceptual tasks. Rather than using luminance noise, which might have degraded the visibility of the target stimulus in the original study by Wagner et al. (2010), motion and orientation noise was added to tasks that required integration of local motion and orientation cues, respectively (Webster et al. 2011; Webster et al. 2012). This ensured that the visibility of the stimulus was not altered. People with migraine showed impaired global form perception, with significantly higher estimates of internal noise that was independent of the level of external noise (i.e. additive noise) (Webster et al. 2012). However, abnormal global motion perception was not accompanied by increased internal noise (Webster et al. 2011). Thus, increased noise cannot account for all of the perceptual deficits observed previously in people with migraine.
1.3.3.2 Altered inhibition
Chronicle and Mulleners (1994) first proposed that migraine might involve a failure of cortical inhibition, as a result of hypoxia due to repeated episodes of aura. Under-
29 inhibition has frequently been attributed to reduced inhibitory neuromodulator levels, namely GABA (Chronicle and Mulleners 1994; Mulleners et al. 2001b; Brighina et al. 2009). In support of this, although levels of GABA in people with migraine have not been assayed directly, there is circumstantial evidence that centrally acting GABA agonists work as prophylactic migraine medications (Hering and Kuritzky 1992).
Inhibition plays a fundamental role in brain function. One of the roles of inhibition is in the tuning of neuronal receptive fields by removing weak inputs, sharpening their specificity for various stimulus characteristics such as orientation (Tsumoto et al. 1979; Morrone and Burr 1986). Under-inhibition in migraine would predict an impairment in orientation discrimination as orientation tuning bandwidth increases. This could explain the significantly elevated orientation discrimination thresholds in both MO and MA groups found by McKendrick et al. (2001) and Tibber et al. (2006). Additionally, in the study by Wilkinson and Crotogino (2000), there was a weak trend for impaired orientation discrimination in people with frequent MA attacks, which appears to be consistent with the original hypothesis by Chronicle and Mulleners (1994) suggesting a cumulative effect of aura over time.
Inhibition is also believed to contribute to visual masking, which refers to the reduced visibility of a target when a masking stimulus is presented before (forward masking) or after (backward masking) the target. Macknik and Livingstone (1998) showed that the reduction in the responses of V1 cells in monkeys due to forward and backward masking is associated with increased inhibition. Palmer et al. (2000) used a backward masking task, where a circular target stimulus was followed by a surround annulus mask at different stimulus onset asynchronies. Participants were asked to identify whether the circular target was complete or missing a portion. The strength of the masking stimulus showed a typical U-shaped function of accuracy (proportion of correct responses) versus stimulus onset asynchrony, where accuracy was markedly reduced when the second stimulus followed the target by 39 ms. It was predicted that people with migraine would be less susceptible to masking effects (i.e. more accurate) if inhibitory function was impaired. This result was evident in the MA
30 but not MO participants. Furthermore, when the same task was conducted in a small group of 3 MA participants who were on prophylactic medication for migraine – valproate, a GABA agonist – performance was normalised to the same level as the control participants (Palmer et al. 2000). However, contrary to these earlier findings, Huang et al. (2006) did not find any difference in the strength of inhibition using a forward and backward masking paradigm, nor any difference in cortical activation using fMRI in people with migraine. Shepherd et al. (2011) later replicated the results of both studies (Palmer et al. 2000; Huang et al. 2006) but also incorporated a baseline measurement, where the target was presented alone without a mask. People with migraine were more sensitive at detecting the target alone compared with controls. Taking this into consideration, the authors found that the difference in performance between unmasked and masked conditions was proportionate across the control, MO, and MA groups (Shepherd et al. 2011).
Inhibition has also been posited to control contrast adaptation, as demonstrated at a single-cell level in cortical neurones of cat and monkey (Albrecht et al. 1984; Sclar et al. 1989; Carandini and Ferster 1997), as well as using fMRI (Gardner et al. 2005) and the VER (Heinrich and Bach 2001). Visual cortical contrast gain control was explored using a masking paradigm by McColl and Wilkinson (2000), where the target stimulus (localised patch of vertically oriented cosine gratings, 3 cycles/degree) was presented alone (unmasked), and then superimposed on a higher contrast masking pattern (larger square of sinusoidal grating oriented 15° relative to the vertical, 3 cycles/degree). Two sets of masking conditions were compared, where the mask and target were presented simultaneously (zero delay) or with a 150 ms delay. It was assumed that in the first condition, there would be insufficient time for contrast adaptation to occur, as rapid contrast gain controls operate in a feedback fashion (Wilson and Kim 1998). On the other hand, in the 150 ms-delay condition, the visual system would undergo rapid contrast gain control to adapt to the contrast of the mask, which was always of higher contrast than the target. It was therefore hypothesised that people with migraine would show a relative reduction in the masking effect due to weakened inhibition. There was no difference between the zero-delay and 150 ms delay conditions, implying a similar strength of inhibition
31 between migraine and non-headache control groups. Nevertheless, there was a trend in the two migraine groups for lower contrast detection thresholds without the presence of a mask, suggestive of hypersensitivity and therefore hyperexcitability.
This collection of studies does not provide clear support for a primary under- inhibition, possibly GABA mediated, in migraine. Rather than thinking of the neural anomaly in migraine as a dysfunction of inhibitory function, it has more recently been proposed that migraine is characterised by a dynamic inability to maintain the balance between excitation and inhibition (Vecchia and Pietrobon 2012). To address this possibility, visual perceptual phenomena that enable assessment of both inhibition and excitation simultaneously have been utilised in studies of migraine, as discussed next.
1.3.3.3 Imbalance between excitation and inhibition
The psychophysical tasks discussed in this section provide a means to test inhibition and excitation in the same paradigm. This, in turn, can help determine whether inhibition or excitation, or both, is driving the altered response in migraine. It must be noted that the neural mechanism/s underlying these effects are still being uncovered and interpretations of the results depends on the validity of current neurophysiological models of the connections involved. Thus, a major assumption is that these visual phenomena represent interactions at a single-cell level.
Wagner et al. (2012) applied the twinkling noise after-effect to test whether people with migraine have enhanced excitability or reduced inhibition. This is a positive after-effect, where observers adapt to a grey patch surrounded by luminance noise and subsequently perceive a prolonged twinkling after-image in the non-stimulated area. The duration of after-image is presumably a result of spontaneous rebound post-inhibitory activity and an indirect measure of long-range cortical inhibition elicited by stimulation of cells in the surround during adaptation (Hardage and Tyler
32 1995). As the people with migraine showed shorter after-effects, the authors proposed a mechanism of weakened inhibition.
In contrast, under-inhibition has not been reported using other visual perceptual paradigms such as binocular rivalry. Binocular rivalry is a specific form of perceptual rivalry that occurs when two different images are presented separately to each eye. The visual system is therefore unable to match the two percepts into a binocular percept; rather, perception is alternated between the two monocular inputs. Simpson (1991) showed that a localised increase in inhibition was responsible for suppression of each monocular image. Wilkinson et al. (2008) presented two grating patterns to each eye, tilted at perpendicular angles to one another. The length of time between alternating percepts was measured. A current model of binocular rivalry (Wilson 2007) assumes that the two eyes take turns to inhibit each other. In people with migraine, it was predicted that a reduction in interocular inhibition would lift the suppressive effect of each eye, resulting in faster switching between the two percepts. Although not statistically significant, there was a trend for the opposite effect – both MO and MA groups showed longer intervals between switches. A later study by McKendrick et al. (2011) confirmed this effect using prolonged binocular viewing to elicit perceptual rivalry. A plaid motion stimulus, consisting of two superimposed gratings, was either perceived as a plaid pattern moving in one coherent direction, or as two gratings sliding over each other in opposite directions. On this visual task, as well as on an analogous auditory perceptual rivalry task, people with migraine showed significantly slower switch rates between the coherent and segregated percepts (McKendrick et al. 2011). Taken together, these findings of slower switch rates is in the opposite direction to what is predicted by under-inhibition in migraine.
Decreased inhibition is inconsistent with findings of stronger after-effects in both aura and non-aura groups (Shepherd 2001; Shepherd et al. 2002; Shepherd 2006). For example, with prolonged viewing of a tilted grating pattern, a person will subsequently perceive the bars of a vertical grating as tilted in the opposite orientation to the adapting display. This is known as the tilt after-effect, which has been attributed to a reduction in the response of cells tuned to the orientation of the
33 adapting stimulus (Maffei et al. 1973). Similar effects are also seen for the motion after-effect, where the stimuli are perceived to be moving in the opposite direction as the drifting grating used for adaptation (Barlow and Hill 1963). The neural mechanism/s underlying after-effects are still unclear, with possible contributions from increased inhibition, decreased excitation, and hyperpolarisation of cell membranes that reduces the likelihood of spiking activity (Carandini and Ferster 1997; Carandini et al. 1998; Sanchez-Vives et al. 2000a; Sanchez-Vives et al. 2000b). Nevertheless, the finding of stronger after-effects is inconsistent with what might be predicted from a lack of inhibition (Shepherd 2001; Shepherd et al. 2002; Shepherd 2006).
In the previous studies of motion after-effects (Shepherd 2001; Shepherd 2006), participants were asked to indicate when they no longer perceived the illusory motion. A less subjective measure of the persistence of after-effects was adopted by Battista et al. (2010). In that study, the test stimulus consisted of two overlapping Gabor patches drifting in opposite directions. Without adaptation, the two patches both contributed equally to the person’s perception; hence, the stimulus appeared to be flickering rather than moving predominantly in one direction. After adaptation to one direction of motion, the patch appeared to move in the opposite direction. To restore the original flickering percept, the contrast of the patch to which the person was adapted needed to be increased. The higher the contrast needed to achieve equal contributions of the two moving Gabors, the greater the strength of adaptation. Contrary to previous findings, the strength of the after-effect was decreased. Nevertheless, the results of Battista et al. (2010) are in agreement with previous studies by Shepherd, which suggest that impaired inhibition does not appear to drive differences in perceptual after-effects in migraine.
The main aim of the study by Battista et al. (2010) was to use the motion stimulus by Tadin et al. (2003) as a perceptual analogue of the antagonistic interactions between centre and surround regions of neuronal receptive fields in extrastriate area V5/MT (Allman et al. 1985). At a single-cell level, the receptive field of a neurone can be considered as its region of excitation. In addition, the firing rate of neurones is
34 decreased or increased by stimulation of the surrounding region, known as surround suppression or facilitation, respectively. It has been shown that the suppressive effects in area V5/MT of macaque visual cortex is largely direction-specific and strongest for large, slow-moving stimuli (Pack et al. 2005). In humans, an analogous suppressive effect is revealed by the reduced ability to perceive motion as the size of a moving stimulus increases (Tadin et al. 2003). Battista et al. (2010) measured performance for a drifting Gabor patch of 0.7° and 5° in diameter. With increasing contrast of the large stimulus, people with migraine required longer presentation durations to discern the direction of motion relative to controls, indicating increased suppression. However, no differences in performance were seen for the small stimulus, suggesting that the altered perceptual responses in people with migraine were not due to an overall reduction in sensitivity.
Increased suppression in migraine was also found by Battista et al. (2011) on a perceptual task linked to centre-surround antagonism in V1 (Zenger-Landolt and Heeger 2003). Battista et al. (2011) found increased suppression in migraine using a drifting version of the Chubb illusion (Chubb et al. 1989), which was a grating stimulus of high-contrast surround (95%) and low-contrast centre (40%). Performance on the static version of the task, however, was normal. These consistent findings of increased suppression in migraine appears to conflict with the idea of hyperexcitability. Our current understanding of centre-surround neurophysiology (Angelucci and Bressloff 2006; Schwabe et al. 2010), depicted in Figure 1.5A, can help explain the paradoxical findings in migraine. Although incompletely understood, the neurophysiology underlying surround suppression at V1 is believed to involve both excitatory and inhibitory interactions, the net effect of which is dependent on stimulus contrast (Figures 1.5C and 1.5D). The suppressive effect arises from intracortical lateral inhibitory connections (Figure 1.5A). In addition, feedback neurones from extrastriate areas, which act via excitatory connections, are proposed to further drive the action of inhibitory interneurones (Figures 1.5A and 1.5B). Cortical hyperexcitability in migraine could therefore result in increased inhibitory effects through excessive excitatory feedback to the lateral inhibitory connections (see Figure 1.5A, blue lines).
35
Figure 1.5 – Model of surround suppression in V1. (A) Current understanding of neurophysiology underlying centre-surround antagonism. The minimum receptive field area (white central circle) is the area of excitation under conditions of high contrast. Under low contrast conditions, the receptive field area increases (light grey annulus). Feedforward connections from lateral geniculate nucleus (LGN) target the receptive field centre (green lines). Suppression arises from the horizontal connections within V1 (red lines). Feedback connections from extrastriate visual areas target the far surround (blue lines). (B) Feedforward (subscript FF) and feedback (subscript FB) connections are believed to be excitatory, whereas local inhibition occurs within V1. (C) Suppression strength as a function of stimulus contrast. (D) Suppression strength as a function of centre contrast. Reprinted from Neuroimage, Volume 52, Schwabe, L., Ichida, J.M., Shushruth, S., Mangapathy, P., Angelucci, A. “Contrast-dependence of surround suppression in macaque V1: Experimental testing of a recurrent network model.” pp 777- 792. Copyright (2010), with permission from Elsevier.
36 Such a model still incorporates the possibility for inhibitory drive to be impaired (as originally proposed to result in hyperexcitability, see for example Palmer et al. 2000), but also demonstrates that there can be concurrent changes to excitatory and inhibitory processes in migraine. This excitatory/inhibitory imbalance could account for the range of seemingly contradictory perceptual responses observed in the migraine literature.
1.3.3.4 Low pre-activation level
Based on combined physiological and pharmacological data, it has been possible to infer the influence of specific medications on brain neuromodulatory status. A psychopharmacological study tested the effects of the hallucinogenic serotonin (5- HT) receptor agonist, psilocybin, in normal observers (Carter et al. 2005). Slower binocular rivalry switch rates were related to reduced levels of psilocybin in a dose dependent manner. While serotonin is unlikely to be the only neuromodulator that influences rivalry rates, the slower switch rates observed in people with migraine noted earlier (Wilkinson et al. 2008; McKendrick et al. 2011) could, at least in part, be influenced by reduced serotonin levels. Abnormalities in serotonin metabolism were first suggested by Sicuteri (1972) as an aetiological factor for interictal anomalies in migraine. Although interictal measures of plasma and platelet levels of serotonin are inconclusive (for a review, see Panconesi 2008), there is some evidence from PET studies that low interictal brain 5-HT synthesis is upregulated at the onset of an attack (Sakai et al. 2008).
Low levels of serotonin have been suggested as reflecting a low basal activity in people with migraine, or ‘low pre-activation level’ (Schoenen 1996). According to the ‘ceiling theory’ (Knott and Irwin 1973), increasing stimulation intensity augments the cortical response to a maximum (i.e. the ‘ceiling’), at which point the response is reduced (i.e. habituation) in normal observers. Habituation is believed to protect the cortex against overstimulation; otherwise, repetitive stimulation places substantial energy and metabolic demands on neurones (Sappey-Marinier et al. 1992; Afra et al.
37 1998; Wang et al. 1999). A low pre-activation level might therefore allow for increasing cortical activity (i.e. lack of habituation) before the ceiling is reached.
In support of a link between habituation and pre-activation states in people with migraine, a finding that is often coexistent with deficient habituation of the PVER is an initial reduction in the response. That is, many investigators report a consistent trend for a reduction in PVER amplitude at the beginning of the continuous recordings (Schoenen et al. 1995; Afra et al. 1998; Wang et al. 1999; Bohotin et al. 2002; Ozkul and Bozlar 2002; Di Clemente et al. 2005; Fumal et al. 2006; Coppola et al. 2007b; Magis et al. 2007b; Coppola et al. 2010a; Coppola et al. 2010b; Coppola et al. 2013), although this finding is not universal (Oelkers et al. 1999; Vigano et al. 2013). Likewise, Hoeffken et al. (2009) demonstrated that the first of two paired- pulse responses recorded in close succession was reduced in amplitude. These results are suggestive of an initial reduction in activity at baseline, presenting an alternative reason underlying cortical hyperexcitability in migraine.
1.4 Potential for pre-cortical dysfunction
Most studies of visual processing in the migraine literature focus on cortical processing, as the brain is implicated in migraine pathophysiology (Schwedt and Dodick 2009; Vecchia and Pietrobon 2012; Pietrobon and Moskowitz 2013) and symptoms such as visual aura are typically homonymous (Schott 2007), indicating a cortical origin. However, a key feature of the visual system is its hierarchical arrangement, whereby processing of visual information at the cortex depends, in part, on input from structures earlier in the visual pathway. Without some measure of pre-cortical visual function, it is not clear whether the presumed cortical vision anomalies that have been discussed thus far in this review might be influenced by visual dysfunction at an earlier site. Studies that have addressed this issue in migraine are described below.
38 1.4.1 Parallel pathways
Consideration of pre-cortical vision anomalies in migraine requires an understanding of the visual processing pathways leading from the retina to the visual cortex. Most of the work investigating these pathways is based on primates; hence, this review will focus on work using monkeys or humans. The pre-cortical pathways are described as parallel, carrying relatively independent information regarding the characteristics of an image, such as its spatial and temporal properties and colour. The segregation into functionally distinct pathways is believed to be a strategy for efficient coding and transfer of different information about the visual world (Livingstone and Hubel 1988).
A schematic of the functional segregation of the primate visual system into two major parallel processing pathways appears in Figure 1.6. The division of labour begins at the retina. Based on morphology, the most common retinal ganglion cells are the midget cells (80%; Figure 1.6, unfilled symbols) and the parasol cells (10%; Figure 1.6, filled symbols). At a given retinal eccentricity, midget cells are, on average, smaller in size with a more compact dendritic arbour than the parasol cells (Watanabe and Rodieck 1989). The remainder of cells consists of at least a dozen other morphologically distinct cell types (Dacey et al. 2003), including wide-field cells such as the small bistratified cells (Rodieck and Watanabe 1993). Retrograde labelling from tracer injections has linked the M and P cell types in the retina to specific layers of the LGN (Perry et al. 1984; Rodieck and Watanabe 1993). The six LGN layers (see Figure 1.6) consist of relatively homogeneous cell populations. Midget ganglion cells project to the four dorsal layers of the LGN, or the ‘parvocellular’ (P) layers, whereas the larger parasol retinal ganglion cell axons terminate in the ‘magnocellular’ (M) layers of the LGN, i.e. the two ventral layers. Hence, midget and parasol cells are also referred to as P-projecting and M-projecting cells. In this thesis, the terms P and M will be used to refer to these anatomical pathways.
39
Figure 1.6 – Functional segregation of the primate visual system into the major M and P pathways. The subdivision of M and P retinal ganglion cells that project to the different layers of the LGN is perpetuated to the primary visual cortex and onto higher visual areas. Reprinted from Science, Volume 240, Livingstone, M. and Hubel, D. “Segregation of form, color, movement, and depth: anatomy, physiology, and perception.” pp 740-749. Reproduced with permission of the American Association for the Advancement of Science in the format reuse in a thesis/dissertation via Copyright Clearance Center.
The M and P pathways show functionally distinct properties. The most common invasive method for identifying functional properties in monkeys has been the use of extracellular recordings of electrical potentials in response to stimulation. The first recordings in macaque by Kaplan and Shapley (1982) were made at the site of retinal ganglion cell axon terminations at distinct LGN layers. These recordings identified two functionally distinct neural populations based on contrast gain (Kaplan and Shapley 1982), with M cells showing a greater slope (higher contrast gain) but saturation at lower contrast levels than P cells. Sustained temporal responses and a steady increase in the response to contrast were characteristics associated with the P pathway. Derrington and Lennie (1984) measured the contrast sensitivity profiles of
40 the M and P pathways to moving gratings by placing micro-electrodes in the P and M laminae of the macaque LGN. P cells had small, colour-opponent receptive fields and hence were tuned to high spatial frequency gratings and had the chromatic properties required for colour vision (Derrington and Lennie 1984). On the other hand, M cells were tuned to higher temporal frequencies nearer 20 Hz (Derrington and Lennie 1984), which implies an important function of the M pathway in the detection of flickering and moving stimuli (Schiller and Malpeli 1978; Kaplan and Shapley 1982). Convergent evidence for different M and P pathway functions has come from disruption of the M and P pathways by chemical ablation of distinct LGN laminae and subsequent measurement of visual performance of macaques. Colour discrimination and sensitivity to higher spatial frequency and lower temporal frequency grating stimuli were severely affected by P pathway lesions (Merigan and Eskin 1986; Merigan et al. 1991), whereas M pathway lesions resulted in loss of sensitivity to achromatic stimuli with low spatial and high temporal characteristics (Maunsell et al. 1990; Merigan and Maunsell 1990).
From the retina, there is evidence from injections of anterograde tracer into the LGN layers that the M and P pathways trace to distinct layers of V1. M cells trace to layers 4Cα of V1, whereas P cells project to layers 4A and 4Cβ of V1 (Hendrickson et al. 1978). A simplistic representation of the two major visual streams from the visual cortex is shown in Figure 1.7. There are direct M pathway projections via V1 to cortical area MT, which is principally involved in motion perception (Livingstone and Hubel 1987) and forms part of the parietal stream (the ‘where’ pathway). On the other hand, the P pathway projects via V1 to V4 and is involved in the ‘what’ pathway, or temporal stream (Ungerleider and Mishkin 1982; Livingstone and Hubel 1987). An alternative view of the functional role of these parallel cortical pathways, espoused by Goodale and Milner (1992), is that the visual coding of information required for the perceptual identification of objects occurs in the ventral (‘what’) pathway, whereas action requires that the location of the object is coded by the parietal (‘where’) pathway. Both of these views, nevertheless, reflect the belief that each stream uses visual information about objects and events differently to enable efficient coding of the visual world.
41 Retrograde tracers injected into the superficial layers of V1 above layer 4 have also identified LGN cells in the intercalated zones between the M and P laminae that stain for the neurochemical markers αCAM kinase and calbindin (Hendry and Yoshioka 1994), which are not expressed by the majority of P and M cells. This prompted investigation of a third ‘koniocellular’ (K) pathway (reviewed by Hendry and Yoshioka 1994; Hendry and Reid 2000; Casagrande et al. 2007). It has been shown in marmoset and macaque monkeys that a heterogeneous group of wide-field retinal ganglion cells that carry signals originating in the short-wavelength sensitive (blue) cones provide input to the K zones of the LGN (Szmajda et al. 2008; Percival et al. 2013). Small bistratified cells (Dacey 1996) and LGN K cells have blue-ON (i.e. short-wavelength sensitive), yellow-OFF colour opponency, which send afferents to the cytochrome oxidase blobs in layer 3 of macaque V1 (Chatterjee and Callaway 2003). Thus, together with the P pathway, the K pathway is believed to form the input to the ‘what’ pathway and is involved in colour discrimination (Figure 1.7).
The invasive studies in monkeys described above are not appropriate for studies in human observers. Rather, psychophysical studies have selected stimuli that are based on the optimal response properties of each pathway. This does not guarantee complete isolation of the pathway of interest, as M, K, and P neural populations are not completely spatially separated anatomically, with the possibility that some LGN cell types might receive more than one input, and that smaller populations of cells that project to the same areas as the M, P, and K cells might have different functional properties (for a review, see Callaway 2005). Nevertheless, studies in people with migraine have demonstrated pre-cortical abnormalities consistent with non-selective dysfunction of the M, P, and K pathways.
Khalil (1991) and Yenice et al. (2007) demonstrated reduced contrast sensitivity in migraine to both high and low spatial frequencies ranging from 0.2 to 18 cycles/degree, which implies M and P pathway dysfunction. In support of this, two pedestal contrast-discrimination tasks, developed by Pokorny and Smith (1997) to test the M and P pathways, revealed the same effect size of dysfunction in people with migraine under both conditions (McKendrick and Badcock 2003). In the first
42 condition, the ‘pulsed-pedestal’ paradigm required adaptation to a steady, spatially homogenous background. A four-square array was then presented, with one square that was brighter or dimmer than the rest, which the person was asked to identify. As the P pathway shows good sensitivity and discrimination to constant displays, it is believed to be chiefly responsible for perceptual responses to the pulsed-pedestal stimulus (Pokorny and Smith 1997). In the second case, the ‘steady-pedestal’ paradigm involved presentation of the four-square stimulus array as the adapting stimulus. Pokorny and Smith (1997) showed that discrimination measured under these conditions demonstrate the temporal and contrast gain characteristics of the M pathway. Later, a modification of the technique of Pokorny and Smith (1997) involving low spatial frequency grating stimuli, as described by Leonova et al. (2003), was applied by McKendrick and Sampson (2009). In that study, contrast discrimination thresholds of migraine sufferers were affected under both steady and pulsed pedestal paradigms, implying that low spatial frequency contrast sensitivity loss does not necessarily imply P cell loss but can also reflect M pathway dysfunction.
Impaired performance on visual tasks that involve blue-yellow opponency has also been demonstrated in people with migraine both foveally (Shepherd 2005; Tibber and Shepherd 2006) and in the peripheral visual field (McKendrick et al. 2002; Yenice et al. 2006; Yenice et al. 2007). Tibber and Shepherd (2006) tested the magnitude of ‘transient tritanopia’ in people with migraine and non-headache controls, which is believed to be mediated at a retinal level (Valeton and Van Norren 1979). Normally, when the eye is adapted to a bright yellow (long-wavelength) light, a temporary reduction in sensitivity to short-wavelength (blue) stimuli occurs after the offset of the adapting display. This apparent suppression of the short-wavelength cones is proposed to arise from inhibition from the recovering long- and medium- wavelength sensitive cones. Both groups showed a reduction in sensitivity to blue targets; however, the magnitude of the reduction was greater in the migraine group (Tibber and Shepherd 2006). This suggests a greater effect of adaptation. Increased effects of adaptation, consistent with a pre-cortical origin, have also been demonstrated in terms of flicker contrast sensitivity by Karanovic et al. (2011) and Thabet et al. (2013).
43
Figure 1.7 – Parallel processing of visual information in primate. Information about different characteristics of a visual stimulus (icons) is transmitted along these parallel pathways from the retina, to V1 and on to higher brain areas. The three pre-cortical pathways (M, K and P) make up the majority of retinal output. Solid lines indicate the major connections between the visual areas (boxes). Reprinted from Neuron, Volume 13, Van Essen, D.C. and Gallant, J.L. “Neural mechanisms of form and motion processing in the primate visual system.” pp 1 – 10. Copyright (1994), with permission from Elsevier.
44 Several studies have reported results that appear to suggest selective neural pathway dysfunction in migraine. Impaired contrast sensitivity to low spatial frequency gratings (< 4 cycles/degree) only (Benedek et al. 2002) or drifting luminance targets against a spatially or temporally modulated background (Coleston et al. 1994; McKendrick et al. 2001), and reduced sensitivity to flickering stimuli at 10 and 20 Hz (Khalil 1991; Coleston and Kennard 1995), implies magnocellular dysfunction. This result could arise in migraine because generalised neural dysfunction is more apparent and manifests earlier in less redundant systems (i.e. the less populated M and K pathways), as proposed by Johnson (1994) to explain pre-cortical losses in glaucoma. Alternatively, flicker is known to induce higher metabolic demands and alter blood flow in both retina (Riva et al. 2001; Riva et al. 2005) and cortex (Pastor et al. 2003). Disrupted neurovascular coupling in migraine (see Figure 1.2, reproduced from Zaletel et al. 2005) could plausibly lead to functional abnormalities that depend on flicker rate.
1.4.2 Concurrent pre-cortical and cortical dysfunction
A concomitant pre-cortical and cortical loss suggests that the cortical deficit might, at least in part, be a downstream effect of earlier dysfunction. This section focuses on the few studies that have considered this possibility in migraine, as described below.
1.4.2.1 Psychophysical thresholds
The preceding discussion of functionally distinct pathways of the visual system is particularly relevant to investigations of possible pre-cortical dysfunction in migraine, because studies have revealed elevated thresholds in people with migraine using specific stimuli that are processed at a cortical level (e.g. global dot motion stimuli, see McKendrick and Badcock 2004c; Antal et al. 2005; Ditchfield et al. 2006; McKendrick et al. 2006a; McKendrick et al. 2006b). Thus, to consider whether these cortical deficits might result from dysfunction earlier in the visual pathway, it makes
45 sense to probe the pre-cortical pathway that is tuned to stimuli of high temporal frequency (i.e. M pathway).
McKendrick et al. (2001) used a battery of tests to measure contrast processing at a cortical and pre-cortical level. The cortical measures were a global dot motion task and orientation discrimination. As both the M and P pre-cortical streams provide inputs to orientation discrimination (Derrington and Henning 1981), the spatiotemporal characteristics of the grating stimuli were chosen to preferentially bias the response to the M and P pathways. Likewise, the measures of pre-cortical contrast sensitivity aimed to differentiate between M and P responses, as contrast sensitivity is known to be severely affected by experimental lesions of the M and P pathways in primates (Merigan and Eskin 1986; Merigan et al. 1991). Neither cortical nor pre-cortical parvocellular measures (using stimuli of 4 cycles/degree) were impaired in people with migraine, as replicated by Tibber et al. (2006). However, pre- cortical contrast sensitivity was impaired when targets were presented against a temporally modulated background (also presumed to measure M function, see Coleston et al. 1994 and Holliday and Ruddock 1983) in the peripheral visual field, consistent with the visual field defects found on flicker perimetry (Medmont perimeter) in these patients. The form of flicker perimetry conducted by McKendrick et al. (2001) involved luminance targets that were presented on top of a luminance pedestal (‘luminance-pedestal’ flicker). Detection thresholds measured with these pedestal stimuli are known to be elevated relative to thresholds measured when the background shows no change in the time-averaged luminance (‘mean-modulated’ flicker) (Anderson and Vingrys 2000) and reflect different mechanisms of flicker processing (Anderson and Vingrys 2001a). Visual field defects were found in response to more than 8 Hz flicker; therefore, these people had apparent M pathway dysfunction in their peripheral visual field. Furthermore, measures of orientation discrimination using a grating with spatial frequency of 0.5 cycles/degree (presumed to bias towards the M system), as well as global motion perception, were abnormal. The majority of migraine participants (67%) showed abnormal performance on the global motion task. A significant proportion (60%) of these participants also showed
46 deficits on the pre-cortical task that was assumed to provide the major input to the global motion task.
1.4.2.2 Visual field sensitivity
Most of the studies reviewed thus far are investigations of foveal vision. However, as mentioned earlier in this chapter, contrast sensitivity deficits have also been found in the peripheral visual field of people with migraine. The pattern of visual field defect also enables localisation of the defect to a cortical or pre-cortical origin, given the anatomy of the visual pathway, as shown in Figure 1.8.
A feature of the visual pathway is the precise retinotopic mapping of retinal to cortical locations. That is, a given location in V1 will correspond to a particular location at the retina, which in turn corresponds to a point in the visual field. Prior to the optic chiasm, the right and left pathways contain monocular information. Due to the partial decussation at the optic chiasm, the visual cortex of the right hemisphere of the brain receives information from the left visual field, and vice versa. Thus, lesions before the optic chiasm result in monocular deficits, whereas a binocular visual field defect that is present in the same half of the visual field in both eyes (i.e. homonymous) implies a lesion in the post-chiasmal visual pathway. It must be noted, however, that a binocular deficit does not preclude the possibility of two separate monocular lesions, but that a binocular and homonymous deficit is highly suggestive of cortical dysfunction. Such homonymous deficits have been found from serendipitous visual field tests during migraine visual aura (Bowerman 1989; Ebner 1991; Luu et al. 2010; Goodwin 2011).
However, the majority of visual field defects found in the interictal period are monocular (for example, see Figure 1.9) (Lewis et al. 1989; Drummond and Anderson 1992; De Natale et al. 1993; McKendrick et al. 2000; McKendrick et al. 2002; McKendrick and Badcock 2004a; McKendrick and Badcock 2004b; McKendrick and Badcock 2004c; Yenice et al. 2006). Patterns of defect range from
47 general depression in one eye, to localised defects and isolated scotomas. Some authors have even described the visual field defects as ‘glaucomatous’ because of their arcuate pattern (Nizankowska et al. 1997; Çomoglu et al. 2003).
Figure 1.8 – The visual pathway. The pairs of circles (labeled 1 to 7) represent the visual field defects observed as a result of characteristic lesions along the visual pathway (denoted as lines corresponding to numbers 1 to 7). Pre-chiasmal defects are monocular, whereas post-chiasmal defects are binocular and homonymous. Reprinted from Snell, R. and Lemp, M. “The Visual Pathway” in Clinical Anatomy of the Eye. pp 379 – 412. Copyright (1998) by Blackwell Science Ltd., with permission from John Wiley and Sons.
48 These observations suggest a pre-cortical origin and are unlikely to be remnants of visual aura, because deficits can occur even in individuals who have never experienced an aura as part of their migraine attacks (Drummond and Anderson 1992; De Natale et al. 1993; McKendrick et al. 2000; McKendrick et al. 2002; Çomoglu et al. 2003; McKendrick and Badcock 2004a; McKendrick and Badcock 2004b; McKendrick and Badcock 2004c). Moreover, both migraine subtypes, MO and MA, display similar patterns of defect.
Visual field assessment has consistently identified deficits in between migraines using standard automated perimetry (SAP) (Lewis et al. 1989; Drummond and Anderson 1992; Wakakura and Ichibe 1992; De Natale et al. 1993; Sullivan-Mee and Bowman 1997; McKendrick et al. 2000; Çomoglu et al. 2003; Goodwin 2011; Dersu et al. 2012), which utilises an achromatic, static stimulus. This stimulus is understood to be non-selective for pre-cortical pathway based on evidence from extracellular recordings of primate retinal ganglion cells showing both M and P cell signatures of contrast gain in response to the SAP stimulus (Swanson et al. 2011).
Perimetric techniques have also been customised to preferentially assess the pre- cortical parallel pathways, such as the fast-flickering luminance-pedestal targets of temporal modulation perimetry (TMP) to test the M pathway (Vingrys et al. 1994) and the blue-on-yellow stimuli of short-wavelength automated perimetry (SWAP) to assay function of the K system (Sample and Weinreb 1990). Visual field defects have been demonstrated in people with migraine using TMP (McKendrick et al. 2000; McKendrick and Badcock 2004a; McKendrick and Badcock 2004b) and SWAP (McKendrick et al. 2002; Yenice et al. 2006). In some instances, fast-flickering and blue-on-yellow test stimuli have identified deficits despite normal visual field results on SAP (Figure 1.9) (McKendrick et al. 2000; McKendrick et al. 2002; McKendrick and Badcock 2004b). McKendrick et al. (2000) additionally demonstrated temporal tuning of the visual field defects in a small group of four migraine sufferers, with deficits becoming apparent at frequencies above 8 Hz. Again, this raises the possibility that techniques that preferentially bias processing to functionally distinct neural pathways are more likely to reveal dysfunction in migraine.
49
Figure 1.9 – Example visual field results in a 27-year-old migraine with aura sufferer, tested 10 days after a migraine event. Shading indicates departures from the control group performance at p < 0.025. At the time of testing, this person had suffered migraines for approximately five years, with an average attack frequency of four per year. The top panels (A) and (C) show the right and left eye results, respectively, in response to static targets, whereas the bottom panels (B) and (D) show the right and left eye results, respectively, in response to temporally modulated targets. From McKendrick, A.M., Vingrys, A.J., Badcock, D.R., and Heywood, J.T. “Visual field losses in subjects with migraine headaches.” Investigative Ophthalmology and Visual Sciences 2000; 41: 1239 – 1247. Reproduced with permission of Association for Research in Vision and Ophthalmology in the format Republish in a thesis/dissertation via Copyright Clearance Center.
In the study by McKendrick and Badcock (2004c), participants completed frequency doubling perimetry (FDP) and motion coherence perimetry (MCP) as a measure of pre-cortical and cortical function, respectively. The FDP test used low spatial frequency grating stimuli (0.25 cycles/degree) that underwent counterphase flicker at high temporal frequency (25 Hz) to preferentially bias the M pathway. The MCP task
50 involved integration of local motion signals; hence, it was presumed to measure extrastriate cortical processing. While no overall difference in migraine group performance was found on FDP, as later replicated by Harle and Evans (2005), seven of the 36 individuals with migraine (19%) showed significant losses of sensitivity on FDP. The number of locations across the visual field that were identified as abnormal on both FDP and MCP tasks was similar, with the exception of one participant (McKendrick and Badcock 2004c). This strongly suggested that, in these individuals, impaired M processing at a pre-cortical level was contributing to the impairment on higher-order motion processing. The strength of the study by McKendrick and Badcock (2004c) was the measurement of visual performance of both eyes, which allowed classification of the spatial extent of MCP deficits as bilateral or monocular. Unlike the seven individuals who showed visual field defects that were consistent with a pre-cortical loss, eight other individuals demonstrated bilateral and homonymous deficits on MCP, which confirmed that these higher- order anomalies were likely to originate from the cortex.
1.4.2.3 Electrophysiological responses
The vast majority of electrophysiological studies in migraine have only measured the cortical response (VER) (for reviews, see Ambrosini et al. 2003 and Magis et al. 2007a). However, the VER also depends on the integrity of the visual pathway from retina to cortex (Fahle and Bach 2006; Odom et al. 2010). Retinal disease can produce abnormal downstream dysfunction in the cortex (e.g. Lennerstrand 1982; Celesia and Kaufman 1985; Holder 1989). Hence, an abnormal cortical response is anatomically non-specific unless earlier visual function is also considered.
In humans, the only neural component of the pre-cortical visual pathway that can be assayed non-invasively using current electrophysiological techniques is the retina, because of the proximity of retinal evoked potentials to the surface of the eye. Retinal ganglion cell activity is evoked by the presentation of patterned stimuli, known as the pattern electroretinogram (PERG) (Maffei and Fiorentini 1981; Sieving
51 and Steinberg 1987; Luo and Frishman 2011). Thus, additional measurement of the PERG at the same time as the PVER provides a more comprehensive view of the function of the visual pathway than using the PVER alone. This is important in differentiating clinical patient groups. For example, Lorenz and Heider (1990) compared groups with purely macular (retinal) disease and another group with tumour-related optic nerve dysfunction. In the latter case of post-retinal dysfunction, all 11 patients showed delayed PVER but a normal PERG, whereas all 10 patients with retinal disease showed delays in both the retinal and cortical responses. The two groups could not distinguished on the basis of the PVER alone.
Simultaneous PERG and PVER recordings have rarely been exploited in the migraine literature. Such recordings do not require extra testing time as the same stimulus is used, and are indistinguishable from those made under separate recording conditions (Matsui et al. 1994). A further advantage of simultaneous recording is the capacity to assay whether the transmission of signal from retina to cortex is altered. For example, under conditions of reduced retinal illumination, both the PERG and PVER may be reduced to a similar extent, but the presence of a normal ‘retinocortical time’ (difference between PERG and PVER peak times) implies faithful transfer of signals along the visual pathway (Froehlich and Kaufman 1991).
Only two studies have thus far considered the possibility for concurrent PERG and PVER abnormalities in people with migraine using transient recordings (< 3 Hz stimulation). Khalil (1991) reported no abnormalities in the PERG, but reduced PVER amplitudes in people with long duration of migraine with aura. Likewise, Shibata et al. (1997) found no abnormalities in the PERG, yet people with aura showed significantly higher PVER amplitudes. Thus, the authors concluded that the abnormality in the cortical evoked response was not due to retinal dysfunction. However, despite studies that have detected abnormal steady-state PVER in people with migraine (Diener et al. 1989; Marrelli et al. 2001; Shibata et al. 2008), the underlying steady-state PERG has not been measured.
52 1.5 Potential for adverse sequelae of migraine
Migraine is generally considered a benign condition, as symptoms are transient and disappear at the end of an attack (International Headache Society 2013). However, in the previous discussion, several authors have posited that some abnormal findings in the interictal period, such as PVER deficits (Kennard et al. 1978; Khalil et al. 2000) or impaired inhibitory function on psychophysical tests (Palmer et al. 2000; Wilkinson and Crotogino 2000), might be a result of repeated episodes of migraine that lead to cumulative damage (Chronicle and Mulleners 1994). The possibility for adverse effects of migraine has motivated studies that consider whether visual function is directly related to the occurrence of a migraine, as follows.
1.5.1 Cortical deficits
The definition of visual aura requires the homonymous visual disturbances to subside, typically within 60 minutes of the onset of migraine headache (International Headache Society 2013). However, there have been more unusual case reports of migraine that do not meet the standard migraine criteria, where resolution of the visual disturbance is prolonged. Some studies have shown complete disappearance of the homonymous defect upon cessation of migraine symptoms some hours later (Bowerman 1989; Luu et al. 2010; Goodwin 2011). However, repeat testing of visual fields has revealed a more prolonged time-course for recovery of visual function in other individuals, from 24 hours (Ebner 1991) to several weeks or months (Sullivan- Mee and Bowman 1997; Razeghinejad et al. 2009), whereas some individuals never fully recover and show persistent, homonymous losses of visual field sensitivity (Rich 1948; Wakakura and Ichibe 1992).
As discussed at the beginning of this chapter, visual auras have been associated with prolonged reductions in cerebral blood flow, namely observed at the visual cortex (Olesen et al. 1981; Lauritzen et al. 1983; Hadjikhani et al. 2001). This has led to a hypothesis that migraine could result in cumulative damage to the brain (Chronicle
53 and Mulleners 1994), which might explain the delayed PVER (Kennard et al. 1978) and lower PVER amplitudes (Khalil et al. 2000) in people with long duration of migraine and visual aura. In support of this hypothesis, cross-sectional studies have shown significant inverse correlations between the volume of grey matter, as measured by voxel-based morphometry and fMRI, and attack frequency (Schmitz et al. 2008; Valfre et al. 2008). A similar relationship between the PVER amplitude and duration of disease was noted by Khalil et al. (2000). However, there are some limitations of correlating migraine characteristics with physiological measures at a single time point. Migraine characteristics are based on self-reporting and subjective recall. A further difficulty is that attack frequency also varies over a person’s lifetime, with the prevalence of migraine declining after approximately the age of 40 (Stewart et al. 1992).
Whether migraine attacks result in cortical ischaemia is debatable. Early studies that used intra-arterial Xenon-133 injections and found a 20 to 25% decrease in regional cerebral blood flow during migraine aura (Olesen et al. 1981; Lauritzen et al. 1983) were possibly confounded by the presence of Compton scatter, which would lead to underestimation of the severity of hypoperfusion. This prompted the idea that the reduction in regional cerebral blood flow at the visual cortex may in fact reach ischaemic levels (Skyhoj Olsen et al. 1987) and potentially cause permanent cerebral damage. Indeed, several population studies have found a relatively higher risk of ischaemic stroke in people with migraine (for meta-analyses, see Etminan et al. 2005 and Schurks et al. 2009), which is independent of other cardiovascular risk factors such as atherosclerotic disease. The increase in absolute risk, however, is only marginal and probably not clinically significant (Schurks et al. 2009). Moreover, more recent neuroimaging techniques such as fMRI and PET, which are not susceptible to Compton’s scatter, have consistently identified only mild hypoperfusion that does not reach the level of ischaemia (for a review, see Schwedt and Dodick 2009).
54 1.5.2 Retinal deficits
The potential for adverse effects of migraine outside of the brain has been investigated using common clinical tests of visual function: visual fields and ocular imaging. The majority of visual field defects found interictally are monocular (Lewis et al. 1989; Drummond and Anderson 1992; De Natale et al. 1993; McKendrick et al. 2000; McKendrick et al. 2002; McKendrick and Badcock 2004a; McKendrick and Badcock 2004b; McKendrick and Badcock 2004c; Yenice et al. 2006), implying a pre- chiasmal origin. One study that measured visual fields at two time-points after migraine revealed localised reductions in sensitivity that were worse the day after an attack (McKendrick and Badcock 2004b). Most defects found by McKendrick and Badcock (2004b) were monocular, with diffuse and localised involvement, raising the possibility for adverse effects at a pre-cortical level.
To explain these possible retinal changes, several investigators have invoked a mechanism of vascular dysfunction (Gasser et al. 1990; Flammer et al. 2001). Although the migraine headache is associated with trigeminovascular activation of the cerebral blood vessels (Moskowitz and Macfarlane 1993; Pietrobon 2005; Dalkara et al. 2006; Vecchia and Pietrobon 2012; Noseda and Burstein 2013), as discussed at the beginning of this chapter, there is also evidence for peripheral vascular abnormalities in people with migraine (for a review, see Tietjen 2009). Migraine is associated with peripheral vasospasms (Zahavi et al. 1984; O'Keeffe et al. 1993; Rose et al. 2004; Flammer et al. 2013), which lead to transient reductions in blood flow, as has been observed in the fingers (Gasser and Flammer 1987; Hegyalijai et al. 1997) and in the retina of migraine sufferers (Killer et al. 2003; Abdul-Rahman et al. 2011). Retinal vasospasm might explain the transient, monocular visual disturbances or loss of sensitivity observed during ‘retinal migraine’ (International Headache Society 2013), although such cases are relatively rare (Hill et al. 2007). It might also explain the numerous case reports of permanent vision loss following retinal vascular occlusions (Galezowski 1882; Graveson 1949; Krapin 1964; Coppeto et al. 1986; Katz 1986; Hykin et al. 1991; Glenn et al. 1992; Inan et al. 1994; Beversdorf et al.
55 1997; Gutteridge et al. 2007). These were diagnoses of exclusion after comprehensive medical and ophthalmological examination that find no explanation, other than migraine, for the vascular event.
A vascular mechanism is also presumed to underlie optic nerve head damage in normal tension glaucoma (NTG) (Broadway and Drance 1998), which is a form of glaucoma that is frequently associated with migraine (Corbett et al. 1985; Phelps and Corbett 1985; Orgul and Flammer 1994; Cursiefen et al. 2000). People with NTG have poor blow flow at the optic nerve head (Plange et al. 2003). Visual field abnormalities in NTG are more likely to progress if there is a concurrent history of migraine (Drance et al. 2001), and NTG patients with focal ischaemic type damage to the optic nerve head have a higher prevalence of migraine (Broadway and Drance 1998). These observations have motivated some investigators to consider the possibility for structural abnormalities of the retina and optic nerve head in people with migraine. Ocular imaging techniques, such as optical coherence tomography (Martinez et al. 2008; Gipponi et al. 2013) and scanning laser polarimetry (Martinez et al. 2009), have demonstrated thinner retinal nerve fibre layers in people with migraine with otherwise healthy eyes, although these findings are not universal (Tan et al. 2005; Moehnke et al. 2008). In addition, Martinez et al. (2009) showed a significant inverse relationship between retinal nerve fibre layer thickness and frequency of attacks, which suggests that these retinal structural changes might be sequelae of migraine.
1.6 Overview of thesis
The previous discussion leads to a number of unanswered questions about interictal vision anomalies in migraine. The discordant findings in the literature raise the possibility that mechanisms other than the predominant model of cortical hyperexcitability might be contributing to abnormal visual function in between migraine attacks. One possibility is that cortical anomalies are a consequence of earlier dysfunction in the visual pathway. A second possibility is that some vision
56 anomalies may reflect adverse sequelae of migraine. Specifically, the following hypotheses will be considered in three experimental chapters of this thesis:
1.6.1 Experiment 1
Retinal and cortical abnormalities in migraine (Chapter 2)
The pattern visual evoked response (PVER) is used to assess neural activity at the visual cortex. In studies of people with migraine, PVER abnormalities are generally interpreted as cortical abnormalities (for reviews, see Ambrosini et al. 2003 and Magis et al. 2007a). However, the possibility that deficits in the PVER may arise from dysfunction occurring earlier in the visual pathway (e.g. at the retina) has been under-explored, despite psychophysical evidence for pre-cortical dysfunction in migraine (e.g. Coleston et al. 1994; McKendrick et al. 2001; McKendrick and Badcock 2003; Shepherd 2005; Tibber and Shepherd 2006; McKendrick and Sampson 2009; Thabet et al. 2013). Thus, the first hypothesis is:
H1A: That people with migraine show concurrent retinal and cortical abnormalities in the electrophysiological response.
In Chapter 2, the hypothesis is tested by measuring the retinal (PERG) and cortical (PVER) electrophysiological responses simultaneously in the same individuals on the same day. This method has been chosen as it can determine whether cortical abnormalities are, at least in part, explained by preceding retinal abnormalities by comparing the magnitude and timing of cortical and pre-cortical dysfunction.
Psychophysical tests of both cortical and pre-cortical visual processing have consistently identified dysfunction in people with migraine using temporally modulated stimuli that is otherwise not measurable with static stimuli (e.g. McKendrick et al. 2000; McKendrick et al. 2002; McKendrick and Badcock 2004b;
57 Antal et al. 2005; Antal et al. 2011; Battista et al. 2011). This raises the potential for temporally tuned deficits in migraine, giving rise to the second hypothesis:
H1B: That abnormalities in the electrophysiological response of people with migraine are more pronounced using stimuli that are modulated at a higher temporal frequency.
Thus, in addition to the standard ‘transient’ response, which is elicited using slower rates of stimulation (< 3 Hz), the novel aspect of the study is the measurement of ‘steady-state’ measures (≥ 4 Hz). Steady-state measures are predicted to reveal dysfunction in people with migraine where transient measures do not.
1.6.2 Experiment 2
Visual changes in the days after migraine (Chapter 3)
Previous studies have documented visual field defects in between migraine attacks, which have been shown to be worse the day after an attack and gradually improve with time (Drummond and Anderson 1992; McKendrick and Badcock 2004b). The majority of these defects are monocular and localised, which have been interpreted as retinal defects. Furthermore, damage to the optic nerve head might be related to the occurrence of migraine attacks (Martinez et al. 2009). Taken together, these observations suggest that retinal dysfunction might occur as a result of migraine. This leads to the first hypothesis of this experiment:
H2A: That measures of retinal function are worse in the days immediately after a migraine.
Loss of visual sensitivity implies abnormal functioning of visual neurones. Some individuals with migraine have shown interictal visual field defects that resemble cortical dysfunction, i.e. bilateral and homonymous (Rich 1948; Ebner 1991;
58 Wakakura and Ichibe 1992; Sullivan-Mee and Bowman 1997; Razeghinejad et al. 2009). Thus, reductions in visual field sensitivity might have physiological correlates in the visual cortex. Thus, the second hypothesis is:
H2B: That measures of cortical function are worse in the days immediately after a migraine.
To test these hypotheses in Chapter 3, the PERG and PVER are used to indirectly measure neural activity in the retina and brain, respectively, on the same day as visual field assessment. Participants are assessed at two different times post-migraine – at least seven days after a migraine, and within six days after a migraine.
1.6.3 Experiment 3
Imbalance between excitation and inhibition in visual processing in migraine (Chapter 4)
One of the theories of migraine is that cortical hyperexcitability results from an imbalance between excitation and inhibition (Vecchia and Pietrobon 2012). Whether excitation or inhibition, or both, is affected in migraine is still unclear. One way to distinguish between effects that are driven by net excitation or inhibition is by measuring perceptual analogues of centre-surround suppression. A stimulus with a low contrast centre and high contrast surround is known to produce an effect of perceptual suppression (Chubb et al. 1989). In the study by Battista et al. (2011), which only used a stimulus of 40% centre and 95% surround, people with migraine showed greater perceptual suppression than controls. However, whether there is suppression or facilitation is based on the net effect of complex inhibitory and excitatory interactions, the balance of which depends on stimulus contrast (DeAngelis et al. 1994; Levitt and Lund 1997; Cavanaugh et al. 2002). Thus, to consider whether both suppression and facilitation are altered in migraine in Chapter
59 4, perceptual suppression will be measured at different contrasts to test the first hypothesis:
H3A: That perceptual surround suppression in people with migraine is increased for centre-surround stimuli of lower contrast centre and higher contrast surround.
Hyperexcitability has been attributed to the increased firing of neurones in visual cortex. To explain the finding of increased surround suppression in migraine, Battista et al. (2011) invoked a model whereby excessive excitatory feedback to inhibitory neurones was responsible for increased lateral inhibition. Hence, cortical neural activity is also expected to show a contrast-dependent pattern of hyperexcitability. That is, the response would be increased at low contrasts, but relatively more attenuated with increasing strength of the inciting stimulus, as the net effect of increased inhibition becomes more apparent. In Chapter 4, to infer cortical neural activity, the PVER will therefore be measured across a range of different contrasts. The second hypothesis of the experiment is:
H3B: That the cortical electrophysiological response in people with migraine is increased for stimuli of lower contrast and decreased for stimuli of higher contrast.
In order to explore the hypotheses presented, perceptual and electrophysiological tests of visual function will be conducted in the same groups of people on the same day. The three experiments are reported in Chapters 2, 3, and 4, with the general discussion and conclusions appearing in Chapter 5.
60 2 | Chapter Two Retinal and cortical abnormalities in migraine
2.1 Introduction
The experiment described in this chapter was designed to determine whether there is concurrent retinal and cortical dysfunction in people with migraine. Most studies of visual processing in people with migraine focus on cortical processing, as migraine is a disorder of the central nervous system (Schwedt and Dodick 2009; Vecchia and Pietrobon 2012). As noted in the previous chapter, pattern electrophysiological studies have frequently identified abnormal cortical activity (PVER) in the interictal period (for reviews, see Ambrosini et al. 2003 and Magis et al. 2007a). An abnormal PVER is typically attributed to dysfunctional cortical processing, but it can also result from anomalous input from earlier stages of the visual pathway. For example, retinal disease, as evidenced by abnormalities in the pattern electroretinogram (PERG), can produce abnormalities in the PVER (Lennerstrand 1982; Celesia and Kaufman 1985; Holder 1989). However, the majority of PVER studies in migraine fail to consider whether the retina is normal, despite the possibility for pre-cortical dysfunction as described in Chapter 1. In order to investigate whether aberrant retinal function occurs concurrently with a cortical abnormality in migraine in this study, the PERG and PVER were measured simultaneously in the same individuals. The advantages of simultaneous PERG and PVER recording and its application in studies of visual function have been discussed in Chapter 1.
There are no visual electrophysiological studies to date that have identified a retinal abnormality in migraine. A number of studies have measured the ‘transient’ PERG (< 3 Hz) in people with migraine, reporting normal responses (Khalil 1991; Moreira
61 Filho and Dantas 1995; Shibata et al. 1997). However, none have measured the ‘steady-state’ PERG (≥ 4 Hz), despite findings of abnormal steady-state PVER (Diener et al. 1989; Marrelli et al. 2001; Shibata et al. 2008). Steady-state responses are presumed to measure similar mechanisms as behavioural measures of temporal processing (King-Smith and Kulikowski 1975). Temporally modulated stimuli have consistently identified differences between migraine and non-headache groups using visual psychophysical tests (Drummond and Anderson 1992; McKendrick et al. 2000; McKendrick et al. 2001; McKendrick and Badcock 2003; McKendrick and Badcock 2004a; McKendrick and Badcock 2004b; McKendrick and Badcock 2004c; Antal et al. 2005; Ditchfield et al. 2006; Shepherd 2006; Battista et al. 2010; Battista et al. 2011; Karanovic et al. 2011; Shepherd et al. 2012; Thabet et al. 2013), including tests of pre-cortical function (Coleston et al. 1994; McKendrick et al. 2001; McKendrick and Badcock 2003; McKendrick and Sampson 2009). Psychophysical studies have found that differences in people with migraine are only apparent when tested with temporally modulated stimuli, as performance is not altered with static stimuli (McKendrick et al. 2000; McKendrick and Badcock 2004b; Battista et al. 2011). McKendrick et al. (2000) additionally found temporal tuning in a small sample of four migraine sufferers with flicker contrast deficits becoming apparent at frequencies above 9 Hz. Temporal tuning has also been demonstrated in flash VER studies, with more pronounced abnormalities at 18 to 20 Hz (Nyrke and Lang 1982; Nyrke et al. 1989; Khalil 1991). Collectively, these findings suggest that visual dysfunction in migraine might depend on temporal frequency, in which case the steady-state electrophysiological response may reveal dysfunction in people with migraine that is less apparent or otherwise not detected by transient measures. Thus, unlike the majority of previous studies in migraine, both transient and steady-state measures were included in the experimental protocol.
This study brings together a range of techniques that have been used previously in isolation to build a more complete picture of how the electrophysiological response is different in people with migraine. The following hypotheses were tested:
62 H1A: That people with migraine show concurrent retinal and cortical abnormalities in the electrophysiological response.
H1B: That abnormalities in the electrophysiological response of people with migraine are more pronounced in response to stimuli that are modulated a higher temporal frequency.
Parts of this chapter, including Figures 2.1 to 2.5 and Tables 2.1 and 2.2 in modified form, have been published in the following journal article (see Appendix A):
Nguyen, B.N., McKendrick, A.M. and Vingrys, A.J. (2012) “Simultaneous retinal and cortical visually evoked electrophysiological responses in between migraine attacks.” Cephalalgia 32: 896 – 907.
2.2 Methods
2.2.1 Participants
2.2.1.1 Human research ethics
Ethics approval was granted by the Human Research Ethics Committee of the University of Melbourne (HREC #0932638) and all study protocols adhered to the tenets of the Declaration of Helsinki. Participants were recruited via an advertisement at the University of Melbourne and online headache support group (Headache Australia) volunteer recruitment webpage, as well as from a database of previous study participants who had consented to having their contact details available for future studies in our research laboratory. All participants were required to read a detailed Plain Language Statement and sign a consent form prior to participation. Testing occurred at the Department of Optometry and Vision Sciences
63 at the University of Melbourne, and participants were reimbursed $20 per test session to assist with any travel expenses incurred from attending.
2.2.1.2 Eligibility criteria
Screening examination
Participants were screened by an optometrist to confirm normal vision and ocular health. Best-corrected visual acuity was required to be 6/7.5 (logMAR) or better with a distance spectacle correction in the range of ±5.00 D sphere and -2.00 D astigmatism, as determined by a subjective refraction. This refractive error criterion was implemented because high refractive errors have been associated with impaired foveal visual sensitivity on psychophysical tests, even though visual acuity is normal (Jaworski et al. 2006). Visual acuity was measured monocularly and binocularly and it was ensured that both of these measures satisfied the inclusion criteria of better than or equal to 6/7.5 (logMAR) to be eligible for the study.
The study was limited to adult participants between the ages of 18 and approximately 45 years, as the prevalence of migraine is highest between the ages of 35 and 45 years, with decreasing frequency thereafter (Stewart et al. 1992). The PERG and PVER are also affected by normal ageing (Celesia et al. 1987). Thus, limiting the study sample to a younger cohort ensured that participants were regular migraine sufferers and minimised the possible confound of age on visual electrophysiological responses. An appropriate near addition was determined for the 50 cm working distance, where required. Previous electrophysiological studies recruited presbyopic participants with normal distance visual acuity (e.g. up to 65 years old, see Shibata et al. 1997), but did not specify whether the volunteers were optimally corrected for the near working distance. Electrophysiological measures are significantly affected by blur (Lovasik and Konietzny 1989; Bach and Mathieu 2004). Light scatter also degrades the retinal image and affects the electrophysiological response (Bach and
64 Mathieu 2004). Slitlamp examination therefore ensured clear ocular media to minimise intra-ocular light scatter.
As migraine has been associated with glaucoma (Corbett et al. 1985; Phelps and Corbett 1985; Orgul and Flammer 1994; Wang et al. 1997; Pradalier et al. 1998; Cursiefen et al. 2000; Hewitt et al. 2010), and because both PERG and PVER measures are affected in glaucoma (Price et al. 1988; Tsaousis et al. 2013), the screening ensured there was no evidence for glaucomatous damage by ocular fundus examination using non-mydriatic ophthalmoscopy. High intraocular pressure (> 21 mmHg), which is a major risk factor for glaucoma, was also ruled out using Goldmann applanation tonometry.
A key characteristic of glaucoma is narrowing of the neuroretinal rim of the optic nerve head with progression of the disease. To obtain a quantitative measure of the amount of neuroretinal rim, the optic nerve head was imaged in 3D using the Heidelberg Retinal Tomograph (HRT, Heidelberg Engineering, Heidelberg, Germany). The Moorfield’s Regression Analysis (MRA) tool of the HRT was used to distinguish between normal and glaucomatous eyes, which has been described in detail elsewhere (Wollstein et al. 1998). In brief, the MRA assumes a linear relationship between log neuroretinal rim area and optic disc area to define the normal ranges. Prediction intervals are calculated around the average linear regression line, taking into account the normal decline in neuroretinal rim area with age. An optic nerve head that is classified as ‘outside normal limits’ falls below the lower 99.9% prediction interval for the group of eyes in the normative database. This classification has excellent specificity (> 85%) for cross-sectional diagnoses of open- angle glaucoma (Wollstein et al. 1998). None of the participants in this experiment was classified as ‘outside normal limits’ using the MRA tool.
All participants were required to have good general health and no systemic disease and/or medications known to affect visual or cognitive function (e.g. antidepressants). Participants who were taking regular prophylactic medications for migraine were also excluded.
65 Classification
Migraine participants were required to meet the criteria of the International Headache Society for migraine without aura (MO) or migraine with aura (MA), as outlined in Chapter 1. At the time of recruitment for this experiment, the previous edition of the International Headache Society classification guidelines was used (i.e. International Headache Society 2004). The diagnosis of migraine was confirmed via clinical interview and a Headache Questionnaire was completed to provide information regarding the characteristics, symptomatology, and treatment of migraine attacks. A full list of the medications used by the migraine participants for pain relief and anti-nausea at the onset of, and during a migraine attack, is given in Appendix A, Table 7.1). A copy of the Headache Questionnaire appears in Appendix A, Section 7.1.2. People with chronic migraine (more than 14 days of migraine per month for more than 3 months, according to the International Headache Society 2004), or who suffered migraines more frequently than once per week, were excluded.
Non-headache control participants also completed an interview and the Headache Questionnaire, verifying that they had never experienced a migraine and did not regularly have headaches (defined as less than four headaches in the past year).
2.2.1.3 Power analysis
The experiment in this chapter was primarily designed to test the first hypothesis (H1A) that people with migraine show abnormal retinal electrophysiological responses. Glaucoma is a condition known to cause retinal ganglion cell loss; hence, it is a useful model in the context of testing for retinal deficits in this study. Specifically, the steady-state PERG shows relatively greater attenuation in glaucoma compared to transient PERG measures (reviewed by Bach and Hoffmann 2008). The power analysis in G*Power (Version 3.1.3, Heinrich Heine Universität, Düsseldorf, Germany) was therefore based on data from studies that have previously used the
66 steady-state PERG to test people with early glaucoma (Trick 1985; Bach and Speidel- Fiaux 1989). The analysis indicated that 17 participants in each group provided a power of 80% for detecting a reduction (α = 0.05) in the steady-state PERG response of one-third the magnitude observed in people with early glaucoma, because more subtle deficits are expected in people with migraine.
In total, forty-five people with migraine were recruited into this study. Of these, 26 people were classified as MO participants (aged 20 – 41, mean ± standard deviation: 28 ± 6 years) and 19 with MA (aged 19 – 43, 33 ± 6 years). Thirty non-headache controls also participated (aged 19 – 46, 26 ± 7 years), who were approximately age- and gender-matched. The proportion of males to females was not significantly different between each group (controls 9: 21, MO 4:22, MA 5:14; chi-square tests of proportions, p > 0.05). There was a greater proportion of female than male participants in each group, as epidemiological reports suggest two to three times the prevalence of migraine in women (Stovner et al. 2007).
2.2.1.4 Participant characteristics
Details of the migraine characteristics of the MO (n = 26) and MA (n = 19) groups are shown in Figure 2.1. Migraine participants also completed the Migraine Disability Assessment Score (MIDAS) Questionnaire to describe the impact of their migraine attacks on activities on daily living (see Appendix A, Section 7.1.2). The MIDAS questionnaire score measures migraine-related disability in terms of days missed and days of lost or significantly reduced productivity at work, school, home, social, and leisure activities (Lipton et al. 2001b; Stewart et al. 2001). The total MIDAS Questionnaire score indicates the level of disability, which is categorised as minimal (score 0-5), mild (score 6-10), moderate (score 11-20), or severe (score 21+). It is an indirect way of estimating the overall severity of migraine attacks, and corresponds well to physician’s judgments of disease severity (Stewart et al. 2001).
67 Participants with MA had a longer duration of migraine history (Figure 2.1C, t43 = 3.24, p = 0.002) and were older than the control and MO groups (Kruskal-Wallis one-way ANOVA, p < 0.001). The MO group reported significantly greater impact of their migraines on daily activities over the past three months as reflected in the higher MIDAS questionnaire scores (Figure 2.1F, Mann-Whitney U = 131.5, p = 0.008). Otherwise, there was no significant difference in migraine characteristics between the MO and MA groups.
Figure 2.1 – Self-reported migraine characteristics of the MO (n = 26, circle symbols) and MA (n = 19, square symbols) participants. (A) Days since last migraine (B) Age at first migraine (C) Years of migraine (D) Number of weeks between migraines (E) Number of lifetime attacks (estimated from number of weeks between migraines and years of migraine) (F) MIDAS questionnaire score (total number of days of lost productivity due to migraine over the past three months). Data from the four participants who were tested one day before a migraine are shown as cross symbols. Horizontal lines indicate the median. Independent sample t-tests and Mann Whitney rank sum tests comparing MO and MA groups are shown. ** denotes significant difference between MO and MA groups at p < 0.05.
68 2.2.1.5 Timing of the test visit
People with migraine were tested at least seven days after a migraine. As changes in PVER amplitude have been reported in the pre-attack period, up to 72 hours before a migraine (Judit et al. 2000; Sand et al. 2008; Sand et al. 2009), all participants were contacted after the test session to determine whether they had experienced a migraine in the three days post-testing. The majority of participants did not suffer a migraine within that time. Four of the 45 migraine participants (9%), however, reported a migraine the day after testing. Data from these participants are represented as cross symbols in all figures and are discussed later.
2.2.2 Pattern electrophysiology
Pattern electrophysiology has been used extensively in studies of migraine; however, despite attempts to standardise tests in this area (Magis et al. 2007a), procedures vary considerably from study to study. In this thesis, a standard clinical approach was taken, based on the recommendations from the International Society for the Clinical Electrophysiology of Vision (ISCEV). The standards define a minimum stimulus and recording protocol for obtaining good quality and reproducible electrophysiological responses (Odom et al. 2010; Bach et al. 2013).
2.2.2.1 Equipment and room setup
The stimuli for pattern electrophysiology were presented on a Sony G520 Trinitron 21-inch cathode ray tube (CRT) monitor (Sony Corporation, Tokyo, Japan) with a resolution of 1024 × 768 pixels and a frame rate of 100 Hz. Gamma corrections were conducted approximately monthly over the course of experimental testing (see Appendix A, Section 7.2). The CRT monitor was turned on at least 30 minutes prior to testing and calibration (Metha et al. 1993). Calibrations were performed using a ColorCal colorimeter with photopic filter (Cambridge Research Systems, Kent, UK),
69 unless otherwise stated. Input to the CRT was controlled by a proprietary computer system (Espion E2 electrodiagnostics system, Version V5, Diagnosys LLC, Cambridge, UK). Room lights were turned off to minimise the effect of scattered light, which is known to reduce the magnitude of the electrophysiological response (Hess and Baker Jr 1984). Two layers of thick, black light-tight curtains were used to section off the recording equipment from the operator’s computer and eliminate any stray light from the computer screen, door, and power switches. The participant was seated comfortably facing the CRT monitor with the head stabilised by a chin-rest. A central cross target of 0.5° diameter was provided to aid fixation.
Pupils were not pharmacologically altered, as pupil dilation results in degradation of the retinal image (Odom et al. 2010; Bach et al. 2013). Consistent with Drummond (1987), there was no difference between control and migraine participants in pupil size prior to electrophysiological recording, as measured with a ruler under dim light (see Appendix A, Section 7.5.1). Moreover, participants wore the appropriate refractive correction for the working distance (50 cm); hence, it was unlikely that fluctuations in pupil size due to changes in accommodation would affect the electrophysiological measures obtained in this experiment.
2.2.2.2 Simultaneous PERG and PVER
To address the first hypothesis (H1A), the PERG and PVER were recorded simultaneously. Responses were recorded monocularly. Although binocular recordings produce larger responses (Heravian-Shandiz et al. 1992) and allows more stable fixation by the better eye (Bach et al. 2013), monocular stimulation enables inter-ocular comparisons to detect possible asymmetry between the left and right eyes. Right eye recordings always preceded recordings from the left eye, to avoid constant stimulation of the same eye and therefore possible adaptation effects.
Electrodes were placed as close as possible to the source of the signal of interest, i.e. on the eye (PERG) and on the back of the head (PVER). These electrodes provided
70 input to the Espion system via five recording ‘channels’ (Figure 2.2A), which were all referenced to a common ground. Each channel comprised one active (positive) and one inactive (negative) electrode (Figure 2.2A), the recorded response being the difference in voltage between the two (Weisinger et al. 1996).
Figure 2.2 – Schematic of electrode placement for simultaneous PERG and PVER recording. (A) The five recording channels, each consisting of an active (positive) and inactive (negative) electrode and referenced to a common ground. (B) The PERG active electrode was a DTL electrode placed along the lower lid margin (left eye shown only). The inactive electrode was a skin electrode placed at the ipsilateral canthus above the cheekbone. (C) The main PVER active electrode was placed along the midline at OZ. All active PVER electrodes were referenced to the inactive electrode at FZ. The common ground electrode for both PERG and PVER was a scalp electrode at CZ. (D) The PVER was also recorded at the right (OR) and left (OL) hemispheres using extra scalp electrodes placed at 10% inion-nasion distance to the right and left of OZ.
71 Ocular electrodes
The PERG active electrode was a DTL carbon-fibre electrode (Roland Consult, Brandenburg, Germany) placed in the conjunctival fornix of the lower eyelid (Figure 2.2B). DTL electrodes are the most comfortable of all electrodes that make contact with the eye and do not require the use of topical anaesthetic (McCulloch et al. 1997). Nevertheless, ocular lubricants (Systane Ultra, Alcon Laboratories, Frenchs Forest, NSW, Australia) were instilled for comfort when required. The corresponding inactive silver-silver chloride skin electrode (Viasys Healthcare, Madison, WI, USA) was placed at the ipsilateral canthus above the cheekbone (Figure 2.2B), after cleansing of the skin with preparation pads containing 70% isopropyl alcohol and pumice (Professional Disposables Inc., Orangeburg, NY, USA) to facilitate attachment and reduce impedance. Placement of the inactive PERG electrode on the outer canthus of the same eye minimises contamination of the signal by responses generated in the other eye or in the cortex, which occur if the ear or forehead is selected as reference (Berninger 1986; Holder 1991).
Scalp electrodes
The PVER was recorded using gold-cup scalp electrodes (Grass Technologies, West Warwick, RI, USA) filled with Ten-20 conductive paste (D.O. Weaver & Co., Aurora, CO, USA). Prior to electrode placement, the skin was prepared with a mild abrasive and moisturising gel (Nuprep skin prep gel, D.O. Weaver & Co., Aurora, CO, USA) to improve conductivity. The electrodes were secured in place with hypoallergenic surgical tape (3M Micropore tape, 3M, North Ryde, NSW, Australia) at specific locations according to the standard International 10-20 system (Jasper 1958; American Clinical Neurophysiology Society 2006). The distance between the two bony protuberances of the head, the nasion and inion (Figure 2.2C), was measured with a tape measure and divided into 10% proportions. The active PVER electrode was placed along the vertical midline over the occipital lobe at Oz at 10% of the nasion-inion distance above the inion (Odom et al. 2010), with the inactive
72 electrode over the frontal lobe at FZ (30% nasion-inion distance above the nasion) according to ISCEV guidelines (Odom et al. 2010). The ground electrode was located halfway between the nasion and inion at CZ (Figure 2.2C).
To consider the possibility for interhemispheric asymmetry in the PVER in people with migraine, as reported in several earlier studies (Tagliati et al. 1995; Shibata et al. 1997; Shibata et al. 1998; Logi et al. 2001; Coppola et al. 2007a), the PVER was also measured at the right (OR) and left (OL) hemispheres (Figure 2.2D). Interhemispheric asymmetry (%) was defined as the absolute difference between the two lateral measurements, relative to the largest signal (Tagliati et al. 1995; Shibata et al. 1997).
Electrode impedance
The ISCEV guidelines recommend that impedance of each electrode be maintained below 5 kOhms (Brigell et al. 2003). This was achieved for all PERG electrodes. However, in some participants, despite multiple attempts to exfoliate the skin of the scalp, the person’s hair prevented good contact between the electrode and the source of the signal. In such cases the impedance was maintained between 5 and 10 kOhms. Thus, in all participants, electrode was impedance was ≤ 10 kOhms. The difference in impedance between active and inactive electrodes was maintained at less than 20% (Brigell et al. 2003).
2.2.2.3 Acquisition settings
Signal acquisition
A total of 200 signal traces were recorded for each stimulus condition. These were not recorded continuously. Rather, in order to minimise the potential for aversion to the patterned stimuli in people with migraine (Marcus and Soso 1989), each stimulus presentation was limited to approximately 20 seconds, corresponding to 25 traces, followed by a brief break of less than 10 seconds. Extraneous bioelectrical signals
73 (e.g. from unrelated muscle activity) and underlying EEG signals that were mutual to both active and inactive electrodes were eliminated by the inbuilt common mode rejection amplifier. Signals were amplified (10 times), bandpass-filtered (second order Bessel filter with cut-offs at 1.25 and 100 Hz), and digitised at a sampling rate of 1000 Hz to 16-bit resolution.
Artefact rejection and signal averaging
Individual traces containing high amplitude artefact (e.g. from blinks and eye movements) exceeding ± 50 µV were removed from the recordings (Odom et al. 2010; Bach et al. 2013). The pre-trigger time was fixed at 12 ms, such that the response was always registered at the same time relative to stimulus onset. This enabled stimulus-locked signal averaging, whereby non-synchronous noise was averaged to zero, thus facilitating extraction of the signal (Weisinger et al. 1996).
Bandpass filter
The bandpass filter cut-off settings were chosen to remove low and high frequency signals and ensure accurate representation of the PERG and PVER waveforms – unwanted DC drift occurs at very low frequencies (Weisinger et al. 1996), whereas pattern evoked responses are slow oscillations, where most of the response energy is confined between 3 and 30 Hz (Brigell et al. 2003). The choice of low-pass filter (100 Hz) was also based on the digitisation rate (1000 Hz), as filter settings exceeding one- fifth of the sampling rate (200 Hz) are known to produce substantial temporal aliasing and low frequency amplification artefact (Brigell et al. 2003).
74 2.2.2.4 Stimulus characteristics and waveform analysis
Contrast
The stimulus was an achromatic square-wave checkerboard, the contrast (%) of which was calculated using Equation 2.1, according to the standard Michelson definition (Michelson 1927):
L − L min max ×100 (Equation 2.1) Lmin + Lmax
where Lmin was the minimum luminance and Lmax was the maximum luminance of a € check. The maximum contrast that could be generated by the CRT monitor was 96 ± 0.6% (see Appendix A, Figure 7.1), which was used in this experiment. € €
Mean luminance
During recording, it is recommended that the mean luminance of the checkerboard remains constant above 50 cd/m2 (Odom et al. 2010; Bach et al. 2013). Despite a gradual reduction (approximately 10%) in maximum luminance of the CRT over the course of 3.5 years, the mean luminance at the centre of the screen remained consistent at 53 cd/m2 and did not vary by more than 1% (see Appendix A, Figure 7.2). As it is known that a CRT monitor does not provide a truly uniform display, mean luminance was also measured at the four corners of the screen. These values were compared to the mean luminance at the centre. Variation in mean luminance from centre to periphery did not exceed 30% (Appendix A, Figure 7.3) in accordance with the ISCEV guidelines (Brigell et al. 2003).
75 Temporal frequency
To address the second hypothesis (H1B), the black and white checks of the checkerboard were alternated at two different temporal frequencies. Firstly, the standard (ISCEV) PERG and PVER response is a ‘transient’ response, elicited using temporal frequencies of less than 3 Hz (6 reversals/second). In this thesis, the ISCEV standard for transient PVER was used, i.e. 1 Hz (2 reversals/second) with an epoch (recording window) of 250 ms (Odom et al. 2010).
On the other hand, there are no ISCEV standards for ‘steady-state’ recording (≥ 4 Hz, 8 reversals/second). Thus, to choose an appropriate reversal rate, several factors were considered. The greatest PERG is elicited using 8 Hz (Hess and Baker Jr 1984; Zapf and Bach 1999; Heine and Meigen 2004), whereas the PVER shows maximal amplitude for temporal frequencies of 4 to 6 Hz (Heine and Meigen 2004). Above 10 Hz, both the PERG and PVER are significantly attenuated (Odom et al. 1982; Heine and Meigen 2004). As a compromise, a reversal rate of 8 Hz was chosen to minimise the duration of each stimulus presentation, as people with migraine can find high- contrast visual patterns aversive (Marcus and Soso 1989). The faster reversal rate allowed more waveforms to be collected in the same amount of time, whilst still producing a response of reasonable magnitude. The exact rate of stimulation was 8.3 Hz (16.7 reversals/second) because the frame rate of the CRT monitor was fixed at 100 Hz, thus limiting the possible temporal frequencies that could be accurately generated with a 10 ms refresh rate. At 8.3 Hz, the checkerboard reversed every 60 ms; hence, one complete cycle from light (ON) to dark (OFF) occurred every 120 ms (see Appendix A, Figure 7.4). Because every change from dark to light represented a single change in contrast, the response elicited occurred at twice the stimulation frequency (Regan 1977), which in this case was at 16.7 Hz, or one complete cycle every 60 ms (see Figure 2.4). Previous work suggests that at least 10 cycles are required to ensure the response reaches steady-state (Armitage et al. 2001); hence, recording was automatically started 750 ms after stimulus onset. An integer number of peaks and troughs were recorded (Bach and Meigen 1999) over an epoch of 480 ms.
76 Spatial frequency
The checkerboard subtended 30°, comprising an even number of black and white squares (checks) of equal size. This ensured that with an even number of pattern reversals over time, the luminance response to light and dark checks cancelled out, leaving only the non-linear response to contrast (Brigell et al. 2003). Check sizes of 0.8° were used, based on the ISCEV recommendation of 1 ± 0.2° (Odom et al. 2010; Bach et al. 2013). In addition, ISCEV standards recommend recordings using smaller (0.25°) checks to confirm the absence of optical problems that may affect the response to very fine elements (Odom et al. 2010).
The steady-state response to 16° checks was also incorporated into the study protocol to enable determination of the steady-state PERG ratio. This was computed by dividing the PERG amplitude in response to 0.8° by the response to 16° checks (Bach and Hoffmann 2008). The PERG ratio shows greater sensitivity (80%) than measures of PERG amplitude alone (70%) for retinal ganglion cell dysfunction in early glaucoma (Bach et al. 2006; Bach and Hoffmann 2008), as the response to large checks is relatively spared. This has been demonstrated in cases of retinal ganglion cell degeneration following optic nerve section in cats (Tobimatsu et al. 1989) and human (Harrison et al. 1987) and in early stages of human glaucoma (Bach et al. 1988). A further advantage of a ratio approach is the reduction in inter-individual variability in PERG measures and the removal of the effect of blur from normal-age related changes (e.g. cataract), which can affect the response to finer checks.
Stimuli of different spatiotemporal frequencies were presented in random order and counterbalanced between participants. A typical recording session lasted approximately 45 to 60 minutes. Data were stored and exported in digital voltage- time format for post-hoc analysis in Microsoft Excel (Microsoft, Redmond, WA, USA).
77 2.2.2.5 Data analysis
Transient waveforms
An unavoidable source of external noise during recordings was the mains power at 50 Hz. This noise appeared as regular high-frequency oscillations in the transient waveforms (see Appendix A, Figure 7.5A). ISCEV advises against notch filtering to specifically remove the 50 Hz component from the waveform (Odom et al. 2010). Instead, a post-hoc centred moving average (sample window: 21 ms) was applied in Microsoft Excel to act as a low-pass filter (see Appendix A, Figure 7.6). Filtering has the inherent limitation of indiscriminate elimination of both signal and noise; thus, regardless of filter cut-off frequency, complete isolation of the signal from noise is not possible. Waveforms were subsequently baseline-corrected by subtracting the signal at 0 ms across the entire waveform, such that all signals were 0 µV at 0 ms.
The average transient PERG and PVER waveforms of the control group (n = 30) are illustrated in Figure 2.3. The methods for extraction and analysis of the different waveform components were as per ISCEV guidelines (Odom et al. 2010; Bach et al. 2013). In all participants, the distinct positive (peaks) and negative (troughs) deflections were identifiable at characteristic ‘peak times’ after stimulus onset. Peak time was measured to the nearest millisecond (ms) from the onset of the stimulus (time = 0 ms) to the positive peak or negative trough (Figure 2.3). Retinocortical time, a measure of signal transmission time between retina and cortex, was calculated as the difference in peak time between the transient PERG P50 and PVER P100 (Kaufman and Celesia 1985). Increased peak time corresponded to a delay in the signal. Peak-to-peak amplitudes were measured in microvolts (µV) from the preceding peak/trough to the component of interest (Figure 2.3). Note that amplitudes measured in this way could not be obtained for the PERG N35 and PVER N75.
78
Figure 2.3 – Control group (n = 30) average transient waveforms and the characterisation of their components according to ISCEV standards. (A) PERG (B) PVER. The PERG was characterised by a positive peak (P50) and two negative troughs (N35 and N95). The PVER was composed of the N75, P100, and N135 components. Peak time (ms) was the duration along the horizontal axis to the peak or trough of the component, relative to the time of stimulus onset (t = 0). The peak-to-peak amplitude (µV) was the height between a successive peak and trough. To illustrate how peak time and amplitudes were measured, the PERG P50 peak time and amplitude are shown in panel (A), whereas panel (B) shows the peak time and amplitude of the PVER P100.
79 Steady-state waveforms
Figure 2.4B shows the average control group (n = 30) steady-state PVER after conversion into the frequency domain by Discrete Fourier Transform (DFT) using the Fourier analysis function in Microsoft Excel. The mathematical computations involved in DFT have been described in more detail elsewhere (Lyons 1997). The DFT requires a binary number of samples in the time series data. Thus, the 480 ms of data were resampled by linear interpolation (Bach and Meigen 1999) to create 512 data points, given that the time between two consecutive data points in the original dataset was 1 ms.
The DFT identified the frequency and magnitude of periodicities in the time series data. As seen in Figure 2.4B, the major response occurred at the contrast reversal frequency, which was the second harmonic (2F, 16.7 Hz). On the other hand, the lack of a significant first harmonic (1F, 8.3 Hz) indicated that the response was not contaminated by luminance artefact (Bach et al. 2013). Analysis in the frequency domain also enabled separation of the signal of interest at 16.7 Hz from the mains interference at 50 Hz
The 2F amplitude (µV) of the complex DFT output at 16.7 Hz was calculated by:
2 2 X(m) = X(m)real + X(m)imag (Equation 2.2)
The 2F phase (π radians) of the complex DFT output at 16.7 Hz was given by: €
y tan-1 /π (Equation 2.3) x
€
80
Figure 2.4 – Control group (n = 30) average steady-state PVER waveform at a stimulation rate of 8.3 Hz (16.7 reversals/second). (A) In the time domain, the response consisted of an integer number (8) of cycles within a recording epoch of 480 ms. A pre- recording epoch of 750 ms was implemented to ensure the response reached steady- state before recording. (B) In the frequency domain, the largest response occurred at the second harmonic (2F, 16.7 Hz). There was comparatively little response at the first harmonic (1F, 8.3 Hz). Noise was estimated from the average of the responses at the neighbouring frequencies around 2F, denoted by vertical dashed lines (14.6 and 18.8 Hz). Note that the mains interference at 50 Hz was adequately separated from the signal of interest at 16.7 Hz in Panel (B).
81 As there could be an infinite set of phases separated by 2 π radians with the same arctangent, phase was constrained to ±1 π radians to avoid cyclic ambiguity. To offset negative phase shifts, phase was reconstructed automatically (unwrapped) by subtracting from a constant of 2 π radians, such that phase was bound between 1 and 3 π radians without discontinuities (Porciatti et al. 2005). The phase convention was the same as for the timing of transient waveforms, such that increased phase values corresponded to signal delays in the time domain.
To determine the significance of steady-state responses, the method described in detail by Meigen and Bach (1999) was adopted. Briefly, every spectral value in the frequency domain consists of both signal and noise components. Using Equation
2.4, the significance ( ps′ ′ ) of a particular signal-to-noise ratio ( s′ ′ ) was calculated by:
1 2 2 2€ 4 2s′ ′ arctan € + 2 2 2 + s′ ′ 2 + s′ ′ ps′ ′ (s′ ′ ) = 2 − 3 (Equation 2.4) 4 + s′ ′ 2 (2 + s′ ′ ) 2
where s′ ′ was the amplitude of the signal of interest (in this study, at 16.7 Hz) and € for the surrounding noise, i.e. the average magnitude at the two neighbouring frequencies (14.6 and 18.8 Hz) (see Figure 2.4B, vertical dashed lines). Steady-state € amplitudes were deemed statistically non-significant if ps′ ′ > 0.05. Using this criterion, the steady-state PVER response of one MA participant was excluded from analysis, while the transient response of this participant was included in the dataset. €
Intrasessional variability
The 200 signals recorded per stimulus condition were divided into two lots of 100. To confirm intrasessional reproducibility, an index of response consistency – the coefficient of variation – was computed between the two consecutive partial averages, using Equation 2.5:
82 σ COV = ×100 (Equation 2.5) µ
where σ was the standard deviation and µ was the mean of the observations from € the same individual. The average coefficients of variation for PERG amplitude (transient: 9%, steady-state: 8%) and PVER amplitude (transient: 11%, steady-state € € 10%) were comparable to those reported in the electrophysiological literature (Bartel et al. 1991; Froehlich and Kaufman 1991; Odom et al. 1992; Porciatti and Ventura 2004; Bowd et al. 2009).
2.2.3 Statistical analysis
2.2.3.1 Group comparisons
Statistical analysis was performed in SPSS version 20.0 (SPSS Inc., Chicago, IL, USA). Data were tested to confirm statistical normality using the Kolmogrov- Smirnov test of normality (p > 0.05) and homogeneity of variances using Mauchly’s test of sphericity (p > 0.05). Where the assumption of sphericity was violated, the degrees of freedom were amended using a Huynh-Feldt correction.
Group comparisons were performed using a repeated-measures analysis of co- variance (RM-ANCOVA) to account for the possible confound of age, with a between factor of group (control, MO, MA) and between factors of eye and component. When the RM-ANCOVA revealed a significant group difference, post- hoc pairwise Bonferroni comparisons were conducted between each possible pair of groups (control versus MO, control versus MA, MO versus MA). As there was no difference between the left and right eyes for all measures (see Appendix A, Section 7.5.2), the right eye response is represented in all figures and tables of this chapter.
83 2.2.3.2 Effect sizes
Effect sizes (Cohen 1988) were determined to enable comparison of the magnitude of deficits between the transient and steady-state measures. The effect size ( d), given by Equation 2.6, represents the mean difference between the control and migraine
(MO or MA) groups, taking into consideration measurement variability.€
d = (µc − µm )/σ pooled (Equation 2.6)
where €
2 2 σ pooled = [(σ c + σ m )/2] (Equation 2.7)
µc was the mean control group amplitude, µm was the mean migraine (MO or MA) € group amplitude, and σ c and σ m were the respective group standard deviations. To construct confidence intervals around the effect sizes, a non-parametric bootstrap € € procedure was used (Efron and Gong 1983). The 2.5th and 97.5th percentiles of effect € € size were determined by resampling with replacement 1000 times from the original
sample of nc and nm observations, where nc and nm were the number of participants in the control (n = 30) and migraine groups (MO: n = 26, MA: n = 19), respectively. € € € €
2.3 Results
2.3.1 The pattern visual evoked response (cortical response)
Table 2.1 summarises the PVER data. PVER timing was not different between groups (RM-ANCOVA main effect of group: F (2, 70) = 0.60, p = 0.55) and will not be discussed further. There was, however, an overall difference between groups in
84 PVER amplitude (main effect of group: F (2, 70) = 3.59, p = 0.03). This difference depended on the PVER component analysed (group x component interaction: Huynh-Feldt ε = 0.93, F (3.74, 130.82) = 3.63, p = 0.009). Hence, the PVER amplitude results are discussed in terms of each component separately.
Of the two transient PVER components (P100 and N135), the P100 amplitude was significantly reduced in the MA group, but not in the MO group, relative to controls (Figure 2.5A; main effect of group: F (2, 71) = 3.67, p = 0.03; post-hoc pairwise Bonferroni tests, control versus MA: p = 0.03). There was no difference between control and migraine groups for N135 amplitude; however, MA participants showed reduced N135 amplitudes compared to the MO group (Figure 2.5B; main effect of group: F (2, 71) = 4.19, p = 0.02; post-hoc pairwise Bonferroni tests, MO versus MA: p = 0.04). Both migraine subgroups demonstrated reduced steady-state PVER amplitudes relative to controls (Figure 2.5C; main effect of group: F (2, 70) = 4.44, p = 0.02; post-hoc pairwise Bonferroni tests, control versus MO: p = 0.01, control versus MA: p = 0.007). Thus, a decrease in PVER amplitude was evident in the transient response of the MA group only, whereas both migraine groups showed reduced steady-state responses relative to controls. The raw waveforms and corresponding Fourier spectra of the 25th, 50th, and 75th percentiles of transient PVER P100 and steady-state 2F amplitude are depicted in Appendix A, Section 7.5. Note that it was not possible to separate migraine and control groups completely by their PVER amplitude, as there was considerable inter-individual variation (Figure 2.5).
Given that the PVER amplitude is altered in the pre-attack period up to 72 hours before a migraine (Sand et al. 2008; Sand et al. 2009), data from the four individuals who reported a migraine the day after testing were excluded (cross symbols in figures). The results of the statistical analyses were unchanged (see Appendix A, Section 7.5.3). The prospect for an interhemispheric asymmetry in the PVER (Tagliati et al. 1995; Shibata et al. 1997; Shibata et al. 1998; Logi et al. 2001; Coppola et al. 2007a) was also tested, by comparing the percentage difference in amplitude between the right-side and left-side measures. Although average interhemispheric
85 asymmetries were higher in the migraine groups, the differences did not reach statistical significance (Table 2.1; main effects of group, p > 0.05)
In accordance with ISCEV standards (Odom et al. 2010), an additional transient PVER measure was obtained using finer checks (0.25°) in conjunction with the standard 0.8° checks. The response to 0.25° checks was within normal limits in terms of amplitude (main effect of group: F (2, 71) = 2.12, p = 0.13) and timing (main effect of group: F (2, 71) = 0.34, p = 0.71) and are presented in Appendix A, Section 7.5.4, confirming the absence of optical problems that may affect the contrast- evoked response to higher spatial frequencies. There was no evidence for an effect of spatial frequency tuning based on the responses elicited by 0.25° and 0.8° checks (interaction between group and check size: F (2, 71) = 0.36, p = 0.70).
Table 2.1 – Summary of PVER amplitude and timing (mean ± standard deviation) and PVER amplitude interhemispheric asymmetry (median, minimum – maximum).
Parameter Controls MO MA p-valuea
Transient P100 amplitude (µV) 11.9 ± 5.3 12.4 ± 5.5 8.0 ± 3.1 ** 0.03
P100 amplitude 15 (1 – 38) 19 (1 – 59) 19 (3 – 85) 0.51 asymmetry (%)
N135 amplitude (µV) 13.4 ± 5.8 16.2 ± 6.1 12.4 ± 4.9 0.02
N135 amplitude 18 (3 – 47) 17 (2 – 67) 17 (5 – 47) 0.90 asymmetry (%)
N75 time (ms) 73 ± 5 71 ± 4 71 ± 8 0.80
P100 time (ms) 102 ± 5 101 ± 4 101 ± 5 0.51
N135 time (ms) 146 ± 15 146 ± 8 142 ± 10 0.57
Steady-state 2F amplitude (µV) 3.7 ± 1.9 2.5 ± 1.4 ** 2.1 ± 1.3 ** 0.02
2F amplitude 16 (1 – 52) 23 (3 – 57) 20 (2 – 39) 0.43 asymmetry (%)
2F phase (π rads) 1.5 ± 0.3 1.5 ± 0.2 1.5 ± 0.2 0.55 a p-values represent the main effect of group (RM-ANCOVA or Kruskal-Wallis one-way ANOVA) ** denotes significant difference relative to controls (post-hoc pairwise Bonferroni tests, p < 0.05)
86
Figure 2.5 – Group mean and individual PVER amplitudes. (A) Transient PVER P100 amplitude (controls: n = 30, MO: n = 26, MA: n = 19) (B) Transient PVER N135 amplitude (controls: n = 30, MO: n = 26, MA: n = 19) (C) Steady-state PVER amplitude (controls: n = 30, MO: n = 26, MA: n = 18). Data from the four participants who were tested one day before a migraine are shown as cross symbols. Error bars represent 95% confidence intervals of the mean. P-values represent significant (p < 0.05) post-hoc Bonferroni comparisons between each possible pair of groups (control versus MO, control versus MA, MO versus MA), given there a significant main effect of group (RM-ANCOVA, p < 0.05).
87 2.3.2 The pattern electroretinogram (retinal response)
Table 2.2 is a summary of the PERG parameters. Although the PVER amplitude was reduced in people with migraine (Figure 2.4), all PERG amplitude parameters were within normal limits (Figure 2.6; main effect of group: F (2, 71) = 0.36, p = 0.70), including the steady-state PERG ratio between responses to large and small checks (Bach and Hoffmann 2008). Neither was there an overall group difference in terms of PERG timing (main effect of group: F (2, 71) = 1.01, p = 0.37). For representative waveforms and corresponding Fourier spectra of the 25th, 50th, and 75th percentiles of PERG amplitude, see Appendix A, Section 7.4.
Table 2.2 – Summary of PERG parameters (mean ± standard deviation).
Parameter Controls MO MA p-valuea
Transient P50 amplitude (µV) 6.6 ± 1.8 6.6 ± 2.1 7.2 ± 2.2 0.45
N95 amplitude (µV) 10.3 ± 2.5 10.6 ± 3.1 10.6 ± 2.8 0.80
N35 time (ms) 23 ± 2 22 ± 3 22 ± 3 0.13
P50 time (ms) 51 ± 2 51 ± 3 51 ± 2 0.26
N95 time (ms) 103 ± 5 105 ± 5 104 ± 5 0.61
Steady-state 2F amplitude (µV) 3.5 ± 0.8 3.3 ± 1.1 3.4 ± 1.0 0.70
2F phase (π rads) 2.1 ± 0.1 2.1 ± 0.1 2.1 ± 0.1 0.26
PERG ratio 1.2 ± 0.2 1.3 ± 0.4 1.2 ± 0.4 0.40 a p-values represent the main effect of group (RM-ANCOVA or Kruskal-Wallis one-way ANOVA)
88
Figure 2.6 – Group mean and individual PERG amplitudes. (A) Transient PERG P50 amplitude (controls: n = 30, MO: n = 26, MA: n = 19) (B) Transient PERG N95 amplitude (controls: n = 30, MO: n = 26, MA: n = 19) (C) Steady-state PERG amplitude (controls: n = 30, MO: n = 26, MA: n = 19). Data from the four participants who were tested one day before a migraine are shown as cross symbols. Error bars represent 95% confidence intervals of the mean.
89 Retinocortical time was similar across groups (controls: 51 ± 4 ms; MO: 51 ± 4 ms; MA: 50 ± 5 ms; main effect of group: F (2, 71) = 0.09, p = 0.91), indicating that the timing of signal transmission between retina and visual cortex was normal in people in migraine. This was unsurprising given that the timing of the PVER (Table 2.1) and PVER (Table 2.2) was normal. On the other hand, there was a trend for a lower ratio between steady-state PVER and PERG amplitudes in people with migraine (controls: 1.14 ± 0.70; MO: 0.90 ± 0.74; MA: 0.70 ± 0.49; main effect of group: F (2, 71) = 3.18, p = 0.05; post-hoc pairwise Bonferroni tests, control versus MO: p = 0.27, control versus MA: p = 0.06), because of the reduction in steady-state PVER amplitude yet normal steady-state PERG amplitudes.
Collectively, the data suggest that the cortical and not retinal response is affected in migraine. Figure 2.7 shows the relationship between the PERG and PVER, by comparing responses that are closest in succession along the visual pathway. That is, for transient responses, the PVER P100 amplitude – believed to reflect an early cortical source in V1 (Di Russo et al. 2005) – was compared to the PERG N95 amplitude. The N95 component was chosen as it is generated almost exclusively by the retinal ganglion cells (Korth 1983; Harrison et al. 1987; Viswanathan et al. 2000) and represents the final retinal output, whereas the P50 component reflects both ganglion cell and pre-ganglionic retinal cell activity (Korth 1983; Luo and Frishman 2011). Similarly, the 2F amplitudes of the PVER and PERG, which measure the responses evoked at V1 (Di Russo et al. 2007) and retinal ganglion cells (Luo and Frishman 2011), respectively, were plotted against one another. Figure 2.7 illustrates that the majority of migraine individuals fell in the lower half of the z-score plots of PVER against PERG amplitude, with the exception of transient P100 amplitudes of the MO participants in Figure 2.7A (left panel).
To explore whether the magnitude of the PERG might contribute to the PVER deficit downstream (H1A), the relationship between the PERG and PVER z-scores, which were normally distributed (Kolmogorov-Smirnov test of normality, p > 0.05), was considered. The PERG did not predict the deficits in the PVER (Figure 2.7 except top-left panel, Pearson correlations, p > 0.05). Where the PERG and PVER
90 amplitudes were normal (Figure 2.7, top-left panel), there was a trend for a linear relationship between the two measures (Pearson r = 0.38, p = 0.05).
Figure 2.7 – PVER amplitude z-scores as a function of PERG amplitude z-scores in the migraine participants. (A) Transient PVER P100 and PERG N95 amplitudes (MO: n = 26, MA: n = 19) (B) Steady-state PVER and PERG 2F amplitudes (MO: n = 26, MA: n = 18). The z-score was determined relative to the control group mean and standard deviation. Data from the four participants who were tested one day before a migraine are shown as cross symbols. Negative z-scores indicate reduced amplitude relative to the mean control group amplitude (horizontal and vertical dotted lines at zero). P-values represent the Pearson correlations between z-scores of the PERG and PVER amplitude measures.
91
2.3.3 Comparison between steady-state and transient responses
To examine whether steady-state electrophysiological measures revealed greater PVER deficits in migraine as per the second hypothesis of this study (H1B), the interaction between group and temporal frequency was considered (RM-ANCOVA, within factor: P100 versus 2F amplitude). This interaction approached statistical significance (F (2, 70) = 2.95, p = 0.06).
The deficits found were also considered in terms of effect sizes. The MO participants showed an average 33% reduction in steady-state PVER amplitude relative to the mean control group steady-state PVER amplitude (Figure 2.5C). The average effect size, d, of this reduction (0.72) was significantly non-zero (95% CI: 0.17 – 1.26), indicating that the reduction in steady-state PVER amplitude in the MO group was€ not due to chance at p < 0.05. The same group of MO participants showed a 5% increase in transient P100 amplitude (Figure 2.5A, unfilled circles). The average effect size of P100 amplitude was d = 0.07 (95% CI: -0.64 – 0.46).
In contrast, for people with aura,€ transient P100 and steady-state PVER amplitudes were, on average, reduced by 32% and 40%, respectively. The effect size of these reductions in amplitude in the MA group was significantly non-zero (transient: d = 0.82, 95% CI: 0.36 – 1.45; steady-state d = 0.90, 95% CI: 0.38 – 1.58). This suggested that, in people without aura, only the steady-state response€ was altered, whereas people with aura showed€ deficits in both the transient and steady-state response. This result was confirmed by the significant difference between control and MO groups in the ratio between PVER P100 and 2F amplitudes (median [range]: controls 3.32 [0.21 – 12.91], MO: 4.55 [1.80 – 11.28], MA: 4.33 [0.37 – 13.39]; Kruskal-Wallis one-way ANOVA: p = 0.04; post-hoc Dunn’s pairwise comparisons, control versus MO: p < 0.05, control versus MA: p > 0.05).
92 2.3.4 Relationship with migraine characteristics
Steady-state PVER amplitudes were reduced in both migraine groups relative to controls (Figure 2.5C). To explore the possibility that this was related to a migraine feature other than the presence or absence of aura, the relationship between the steady-state PVER amplitude (averaged across both eyes) and self-reported migraine characteristics shown in Figure 2.1 was considered. None of the correlations was significant (Table 2.3). People with aura showed an additional deficit in the transient PVER, which was not apparent in the MO group. However, there was no relationship between the reduction in the transient PVER amplitude and the features (i.e. age, years of migraine, MIDAS questionnaire score) that were significantly different between the MO and MA groups in this study (Table 2.4).
Table 2.3 – Relationship between steady-state PVER amplitude and self-reported migraine characteristics of the pooled migraine group (n = 45).
a Migraine characteristic R p-value
Days since last migraine -0.18 0.23
Age at first migraine (years) -0.23 0.13
Years of migraine 0.23 0.13
Weeks between migraines -0.10 0.52
Estimated number of lifetime attacks 0.13 0.39
MIDAS questionnaire score (days) -0.14 0.37 a p-values represent the significance of Spearman rank correlations
Table 2.4 – Relationship between transient PVER P100 amplitude and the features of the MA group (n = 19) that were significantly different from the MO group.
a Migraine characteristic R p-value
Age (years) -0.37 0.12
Years of migraine -0.36 0.13
MIDAS questionnaire score (days) -0.04 0.89 a p-values represent the significance of Spearman rank correlations
93 2.4 Discussion
In this chapter, the following hypotheses were considered:
H1A: That people with migraine show concurrent retinal and cortical abnormalities in the electrophysiological response.
This study involved comprehensive measurement of a range of PERG and PVER parameters. The steady-state PVER was abnormal in both migraine groups (Figure 2.5C; p = 0.02), yet the corresponding PERG was within normal limits (Figure 2.6C; p = 0.70). Likewise, the transient PVER amplitude was reduced in the MA group relative to controls (Figure 2.5A; p = 0.03); however, there was no difference between groups for any of the transient PERG amplitude measures (Table 2.2; p > 0.05). These results do not support the first hypothesis (H1A) of concurrent retinal and cortical electrophysiological abnormalities in people with migraine. Moreover, the amplitude of the PERG did not predict the amplitude of the PVER (Figure 2.7, Spearman rank correlations: p > 0.05), which implies that the PVER abnormality did not arise downstream from retinal dysfunction.
H1B: That abnormalities in the electrophysiological response of people with migraine are more pronounced in response to stimuli that are modulated a higher temporal frequency.
The literature suggests that deficits in migraine might depend on the temporal frequency of visual stimulation. In this study, transient (1 Hz) and steady-state (8.3 Hz) responses were compared (H1B). There was a trend for greater dysfunction in the steady-state than transient PVER (interaction: p = 0.06). This trend was apparent in the MO group, whose average steady-state PVER amplitude showed a significant reduction relative to the control group (33% reduction, d = 0.72, 95% CI: 0.17 – 1.26), whereas the transient response was normal (5% increase, d = 0.07, 95% CI: -
0.64 – 0.46). In contrast, the MA group showed€ significantly reduced PVER €
94 amplitudes for both transient (32% reduction, d = 0.82, 95% CI: 0.36 – 1.45) and steady-state measures (40% reduction, d = 0.90, 95% CI: 0.38 – 1.58). Thus, this study does not provide conclusive € evidence in support of the second hypothesis (H1B) for a more pronounced€ deficit in the steady-state response measured at 8.3 Hz, compared to the transient response measured at 1 Hz.
The nature of the electrophysiological deficits observed in this chapter, and their possible underlying mechanisms, is reflected on in the following discussion.
2.4.1 Retinal electrophysiological responses in migraine
The PERG measures in the migraine participants were not different from controls (Table 2.2), which implies there is no diffuse retinal abnormality in migraine. Previous studies that have measured the transient PERG also support this conclusion (Khalil 1991; Moreira Filho and Dantas 1995; Shibata et al. 1997). A novel addition to the migraine literature is the present finding of a normal steady- state PERG (Figure 2.6C). Despite greater sensitivity in identifying early retinal dysfunction in conditions such as glaucoma (Bach and Hoffmann 2008), the steady- state PERG did not reveal a deficit in people with migraine (Table 2.2).
As there is other evidence for pre-cortical, and possibly retinal, dysfunction in people with migraine (see Chapter 1), the findings of this study do not preclude retinal involvement in migraine. Rather, the electrophysiological techniques employed may be insensitive to the type of dysfunction present in migraine. Unlike glaucoma, migraine is not a predominant and global retinal ganglion cell dysfunction, for which the PERG is a sensitive indicator (Bach and Hoffmann 2008). If pre-cortical dysfunction in migraine is less diffuse in nature, it may not be detected by the checkerboard stimulus used here, which covered 31° of the visual field. Future investigations could make use of multifocal electrophysiological techniques (Klistorner et al. 1998) that are designed to investigate spatially localised defects using smaller targets presented at different locations across the visual field.
95 Furthermore, visual field defects are most prominent the day after migraine (Drummond and Anderson 1992; McKendrick and Badcock 2004b) and there is some evidence for retinal vasospastic events that occur during a migraine attack (Killer et al. 2003), which suggests that time post-migraine might have some bearing on retinal evoked responses. These possibilities are considered in the next experiment (Chapter 3).
Consistent with previous simultaneous recordings of the PERG and PVER (Khalil 1991; Shibata et al. 1997), this study did not find an abnormality in the PERG, yet the PVER was significantly affected in people with migraine. Although these findings demonstrate that a retinal deficit is unlikely to underlie the cortical deficit in migraine, the origin of the abnormal PVER remains uncertain. Given that retinocortical time was normal, this implies that the defects measured are likely to arise from processing anomalies at the visual cortex and not earlier in the visual pathway. However, the PVER also reflects the patency of input of lateral connections within visual cortex and feedback connections from higher cortical areas (Fahle and Bach 2006). The PVER represents a massed population response from many classes of neurones in multiple cortical layers and areas (Di Russo et al. 2005; Di Russo et al. 2007), and cannot be separated into anatomically distinct cell pathways without pharmacological intervention or invasive techniques, as has been performed in rodents (e.g. Tsai 2012). Thus, it remains to be determined which of the many contributions to the PVER is dysfunctional in migraine.
2.4.2 Cortical electrophysiological responses in migraine
Unlike many previous electrophysiological studies in migraine (Ambrosini et al. 2003; Magis et al. 2007a), this experiment explicitly adhered closely to the most up-to-date ISCEV standards for clinical visual electrophysiology (Odom et al. 2010; Bach et al. 2013). Differences in experimental protocols add to the difficulty in reconciling the findings of this experiment with previous literature. Nevertheless, the following
96 discussion attempts to compare the present results with those obtained under similar stimulus conditions by other investigators.
The transient PVER is the standard ISCEV response (Odom et al. 2010) and is most commonly reported in the migraine literature (Ambrosini et al. 2003; Magis et al. 2007a). In this study, a temporal frequency of 1 Hz was used to elicit the transient response. There was a significant reduction (33%) in P100 amplitude in people with aura that was not observed in the MO group (Figure 2.5A). Using the same temporal frequency, Boylu et al. (2010) found reduced P100 amplitudes in 41 people with migraine, but did not distinguish between MO and MA groups. Thus, it is unclear whether the approximate 17% reduction in P100 in their pooled migraine group was driven by a particular subtype, as is suggested by the data presented in this chapter.
Khalil and colleagues considered MO and MA groups separately and found a reduction in P100 amplitude (approximately 20%) in the latter group only (Khalil 1991; Khalil et al. 2000), consistent with Figure 2.5A. However, the reduction was only significant relative to controls for a subgroup of participants with a history of at least 30 years of migraine with aura. For MA participants with a history of less than 10 years, there was an increase in P100 amplitude (Khalil 1991; Khalil et al. 2000). These findings could not be replicated in this study, as the aim was not to recruit participants with varying durations of migraine history. In the MA group, three participants had suffered migraines for less than 10 years, and only one participant for 30 years (Figure 2.1C). Moreover, it was unlikely that people with very long migraine history, and therefore likely to be older, would have been recruited. The age range for inclusion in the study was limited (less than approximately 45 years) to minimise the possible confound of age, which is known to affect visually evoked responses (Celesia et al. 1987).
Other electrophysiological studies in migraine using 1 Hz to elicit transient responses have shown equivocal results, with no difference in P100 amplitude between migraine and non-headache groups (Logi et al. 2001; Marrelli et al. 2001). There are also reports of the opposite finding – an increase in P100 amplitude for people with
97 specific migraine characteristics, such as left-sided headache (Kennard et al. 1978). Heterogeneity of migraine participants may explain these dissimilar results. As demonstrated in Figure 2.1, the people with migraine in this study reported a range of migraine characteristics, which appeared unrelated to the magnitude of the cortically evoked response (Table 2.3). Variability in migraine characteristics is further discussed in the final chapter of this thesis.
Compared to standard transient recordings, there are fewer studies of the steady- state PVER in migraine, and these report higher steady-state amplitudes (Diener et al. 1989; Marrelli et al. 2001; Shibata et al. 2008). Marrelli et al. (2001) tested children with migraine and found higher amplitudes for the first harmonic response only. The presence of a significant first harmonic, however, may indicate technical problems (Bach et al. 2013), as the prominent response to contrast is expected to occur at the second harmonic (Regan 1977). Shibata et al. (2008) observed a significant increase in the amplitude in both MO and MA groups using a checkerboard stimulus consisting of checks of 0.5 cycles/degree spatial frequency (equivalent to 1.4° checks) reversing at 5 Hz. On the other hand, the PVER was normal in response to stimuli of 2 cycles/degree (0.4° checks) reversing at 10 Hz (Shibata et al. 2008), which suggests that dysfunction in migraine might vary with spatial and temporal frequency. The protocol used by Diener et al. (1989) was most similar to the current study in terms of spatial and temporal frequency (1° checks at 8.33 Hz); however, the signals were not analysed in the frequency domain using DFT. Rather, amplitude was measured as the distance between peaks to troughs. This method of amplitude measurement does not provide an accurate reflection of the magnitude of the evoked response (Regan 1982), as superimposition of non-physiological noise (i.e. 50 Hz mains interference) can artificially increase the apparent peak-to-trough distance by introducing oscillations in the waveform (see Appendix A, Figure 7.5A). Variation in test protocols, as discussed more generally in the final chapter (Chapter 5), may explain the discrepant findings.
98 2.4.3 Possible mechanisms for reduced cortical responses
Many previous reports in the migraine literature show an increase in the cortical response (Ambrosini et al. 2003; Magis et al. 2007a), which has been interpreted as evidence for cortical hyperexcitability in migraine (Aurora and Wilkinson 2007; Coppola et al. 2007c), as discussed in Chapter 1. The PVER abnormality found in this study was a reduction in the response without an effect on timing. Although the mechanism for reduced PVER amplitudes in migraine cannot be ascertained from this study, there are several possibilities, which are outlined below.
With the exception of habituation studies (for a review, see Coppola et al. 2009), very few PVER studies specify whether the recordings are made continuously. Without breaks between presentations, increased amplitude may result from the averaging of progressively larger amplitudes over time due to deficient habituation in migraine (Schoenen et al. 1995). The recordings made here used stimuli that were presented for a short time (approximately 20 seconds) after a 750 ms conditioning period, with brief breaks of less than 10 seconds between each presentation. This method is typical of clinical electrophysiological recordings to prevent fatigue and, in this instance, was adopted to minimise the possible influence of habituation. Thus, reduced PVER amplitude in this study might reflect a ‘low cortical pre-activation level’ in migraine (Schoenen et al. 1995), consistent with other electrophysiological evidence for an initial reduction in cortical activity in migraine (see Chapter 1). For example, the PVER amplitude is reduced in the first ‘block’ of responses in many studies of habituation in migraine (Schoenen et al. 1995; Afra et al. 1998; Wang et al. 1999; Bohotin et al. 2002; Ozkul and Bozlar 2002; Di Clemente et al. 2005; Fumal et al. 2006; Coppola et al. 2007b; Magis et al. 2007b; Coppola et al. 2010a; Coppola et al. 2010b; Coppola et al. 2013), although this finding is not universal (Oelkers et al. 1999; Vigano et al. 2013). Similarly, an initially reduced response occurs when two paired-pulse PVER responses are measured in close succession (Hoeffken et al. 2009).
99 Another potential reason for a reduction in the cortical evoked response is an imbalance between cortical inhibition and excitation (Vecchia and Pietrobon 2012), for which there is evidence from perceptual studies of centre-surround suppression (see Chapter 1). For example, a well-studied phenomenon is the Chubb Illusion, or contrast-contrast stimulus (Chubb et al. 1989). The perceived contrast of a central patch is reduced in the presence of a high contrast surround (Cannon and Fullenkamp 1991; Xing and Heeger 2001; Yu et al. 2001). Using a surround of 95% contrast, Battista and colleagues demonstrated a greater reduction in the perceived contrast of a central patch (40% contrast) in people with migraine, i.e. an increase in perceptual surround suppression (Battista et al. 2011). Accordingly, the checkerboard stimulus (96% contrast) used in this study, which covered 30° of the central visual field, may have induced a stronger suppressive effect in the migraine participants. As the PVER is largely driven by the response at the macula (Sakaue et al. 1990), a reduction in the perceived contrast of the centre of the checkerboard stimulus leads to a dampening of the cortical response (Valberg et al. 1981). The possibility for centre-surround effects influencing the PVER amplitude is explored in the final experiment (Chapter 4).
Alternatively, a reduced response might reflect structural damage. Khalil et al. (2000) found a reduction in PVER amplitude in people with a long history of migraine, and hypothesised that damage could be a cumulative result of repetitive ischaemia with migraine attacks. As noted in Chapter 1, however, there is some debate in the literature as to whether cerebral blood flow during migraine actually reduces to ischaemic levels. To consider the possibility for structural lesions in the brain, neuroimaging techniques could be implemented to identify cortical lesions associated with migraine (Schwedt and Dodick 2009). Alternatively, an indirect way to infer whether repeated migraine episodes are associated with cumulative injury is by correlating years of migraine or attack frequency with the PVER deficit, which in this study was not significant (Table 2.3). However, migraine frequency typically varies over a lifetime, with peak prevalence rates among the working-age demographic, i.e. 25 to 55 years (Bigal and Lipton 2009). To determine whether a reduction in the
100 PVER is a result of repeated migraine attacks would require longitudinal investigation, which has not yet been attempted in people with migraine.
It is worth considering whether the results are explicable by non-visual mechanisms. Prolonged testing can induce drowsiness and an inability to maintain attention, which can reduce the PVER (Douthwaite and Connor 1989; Mezer et al. 2004). Although attention was not specifically controlled in this study, there was no reason to suspect that the people with migraine were more drowsy or fatigued than the control participants. Participants were frequently asked during the testing to indicate whether the fixation target was clear, and the stimuli were interleaved with regular breaks. Test sessions were scheduled at least seven days after migraine offset to minimise migraine-related fatigue and medication effects. Moreover, the response to patterns with very small elements (0.25° checks) was not different between groups (see Appendix A, Section 7.5.4), which argues against greater accommodative fatigue in the migraine group.
Similarly, greater aversion to the flickering checkerboard stimulus in the migraine group (Marcus and Soso 1989; Khalil 1991; Coleston and Kennard 1993; Shepherd 2000; Huang et al. 2003; Wilkinson et al. 2008; Karanovic et al. 2011; Thabet et al. 2013) might decrease the ability to voluntarily attend to the target and affect the PVER amplitude (Douthwaite and Connor 1989). Aversion was not formally tested in this study; nevertheless, none of the participants reported discomfort during the recordings or needed to abort the testing due to heightened aversion or an impending migraine attack. Furthermore, participants self-selected to be involved in a study that explicitly involved the use of flickering stimuli. People with strong aversion to visual stimuli may have been less likely to volunteer.
2.4.4 Greater dysfunction revealed by steady-state measures
To address the second hypothesis (H1B) that greater dysfunction would be revealed using steady-state rather than transient measures, the interaction between group and
101 temporal frequency was considered for the PVER amplitude measures (p = 0.06). The lack of statistical significance at p < 0.05 might be because the sample size arrived at by the power analysis was based on the primary hypothesis (H1A) that people with migraine would show deficits in the PERG.
That the MO participants only showed deficits in the steady-state response is consistent with previous literature demonstrating abnormal performance using flickering or moving, rather than static, stimuli (McKendrick et al. 2000; McKendrick and Badcock 2004b; Battista et al. 2011). The difference in transient and steady-state measures between migraine subgroups, however, was an unexpected finding, given the uncertainty in whether MO and MA are different entities (Manzoni and Torelli 2008; Purdy 2008). Further experiments are required to determine whether this difference might be related to a difference in temporal tuning of the cortical deficit. Using flicker perimetry, McKendrick et al. (2000) tested the contrast sensitivity of a small sample of migraine participants (1 MO, 3 MA) and reported greater deficits at frequencies above 9 Hz. Khalil (1991) measured repetitive flash VER at a range of temporal frequencies (6 to 24 Hz) in a group of people with migraine and found the greatest difference occurred at 6 Hz, 18 Hz, and 20 Hz. These observations demonstrate that visual deficits in migraine might show temporal tuning, which warrants future work that is specifically designed to assess the PVER in response to a range of temporal frequencies.
In comparison to the MO group, the MA group showed reduced transient PVER amplitudes (Figure 2.5A), consistent with other studies (Khalil 1991; Khalil et al. 2000; Omland et al. 2013). There is considerable debate in the literature regarding whether MO and MA represent two distinct entities and have different pathophysiological mechanisms, as highlighted in Chapter 1. The fact that people who are categorised as having MA by the IHS classifications (International Headache Society 2013) do not always suffer migraines with aura, and that people can change from having MO to MA attacks within their lifetime (Rasmussen and Olesen 1992), adds further complexity. Nevertheless, one difference between the two migraine groups in this study was the lower MIDAS questionnaire scores reported by the MA
102 participants (Figure 2.1F), which implies less disability due to migraine (Stewart et al. 2001). Less disability, however, is counter-intuitive to the idea that migraine severity might relate to a worsening of the cortical deficit.
Another point of difference was that the MA participants of this study were older than the MO participants, with significantly longer migraine history (Figure 2.1C), thus possibly representing a more severe form of migraine. In people with aura, Khalil et al. (2000) observed that the P100 amplitude declined in proportion to the years of migraine. The vast majority of people with MA experience aura that is visual in nature (Russell and Olesen 1996) and clearly involves the visual cortex. Although highly speculative, reduced PVER amplitude might be an acquired phenomenon due to cumulative ischaemic damage to the visual cortex from repeated migraine attacks with visual aura (Kennard et al. 1978; Khalil et al. 2000). The presence of aura might therefore influence a person’s susceptibility to adverse vascular effects. People who experience aura are at slightly greater risk of subclinical brain lesion, stroke, cardiovascular disease and other ischaemic vascular events than are those who do not (for reviews, see Bigal and Lipton 2009 and Schwedt and Dodick 2009); however, the absolute risk is only marginally higher (Schurks et al. 2009). A larger population study that includes neuroimaging measures would be required to investigate the presence of brain lesions and the effect on the PVER in people with migraine. Alternatively, if damage is related to the migraine event itself, then subtle changes in the PVER may be evident in the days immediately after each attack. The next experiment (Chapter 3) will attempt to examine whether these changes occur.
It is important to note that although neurophysiological testing, such as measurement of the PVER, is a valuable tool for investigating neural function in people with migraine, the experimental methods applied in this study have limited value for migraine diagnosis. Standardisation of methodology is one approach to optimise clinical protocols, as proposed by Magis et al. (2007a); however, it is apparent from this experiment that even with close adherence to the ISCEV standards (Odom et al. 2010; Bach et al. 2013), there is considerable inter-individual variability in electrophysiological data and substantial overlap in outcomes between
103 groups. Future work, including the subsequent experiments presented in this thesis, would therefore benefit from within-individual comparisons and normalisation to address the heterogeneity of PVER findings in migraine.
2.5 Conclusions
In summary, this study extends our knowledge of visual electrophysiological responses between migraine events, by concurrent measurement of retinal and cortical evoked responses at two different temporal frequencies (1 Hz and 8.3 Hz). Similar to previous reports, the PVER was abnormal in migraine sufferers; however, the overlap between the migraine and control groups indicates that these tests are unlikely to be of use as clinical markers of the condition.
The reduction in the cortical response could not be explained by impairments in the eye or transmission along the visual pathway, as the PERG and retinocortical time were normal, at least seven days after a migraine. This is in contrast to the deficits that manifest on visual field testing, which often appear localised and retinal in origin, particularly the day after a migraine. Whether generalised and localised forms of visual deficit are present in the same individuals with migraine is considered in the next experiment (Chapter 3). The next experiment also considers whether, like visual field deficits, electrophysiological deficits might be worse immediately following a migraine event.
Participants who experience migraine without aura showed a reduction in the steady- state but not transient PVER, suggesting greater dysfunction is revealed with steady- state electrophysiological recording. On the other hand, all migraine participants, on average, showed a reduction in the steady-state response. As discussed in this chapter, there are several possibilities as to why the cortical response is reduced in migraine, one of which – a cortical imbalance between inhibition and excitation – is explored further in the final experiment (Chapter 4).
104 3 | Chapter Three Visual changes in the days after migraine
3.1 Introduction
Previous literature does not suggest a single, common anatomical locus for visual deficits in migraine. Chapter 1 described the convergent evidence for abnormalities in both primary visual cortex (V1) and extrastriate visual areas from studies using brain neuroimaging, direct stimulation, electrophysiology, and psychophysical methods. In the first experiment (Chapter 2), the possibility for concurrent retinal and cortical visual dysfunction was explored using electrophysiological techniques. The data suggested cortical involvement only, as abnormal cortical visual evoked responses to patterned stimulation (PVER) occurred concurrently with normal retinal responses (PERG). The findings confirmed involvement of the visual cortex in the pathophysiology of migraine.
However, as outlined in Chapter 1, the majority of interictal visual field findings suggest dysfunction of pre-cortical visual pathways, which appears inconsistent with the electrophysiological findings. Visual field tests are commonly used to identify visual dysfunction due to ocular and neurological disease. Visual field assessment has consistently identified contrast sensitivity losses in between migraine attacks using static (Lewis et al. 1989; Drummond and Anderson 1992; Wakakura and Ichibe 1992; De Natale et al. 1993; Sullivan-Mee and Bowman 1997; McKendrick et al. 2000; Çomoglu et al. 2003; Goodwin 2011; Dersu et al. 2012), flickering (McKendrick et al. 2000; McKendrick and Badcock 2004a; McKendrick and Badcock 2004b), and blue- on-yellow test stimuli (McKendrick et al. 2002; Yenice et al. 2006).
105 Unlike the electrophysiological techniques adopted in the first experiment (Chapter 2), visual field tests provide additional information about whether deficits are diffuse or localised by identifying which locations across the visual field are affected. Two main patterns of visual field loss have been identified in people with migraine: decreased generalised sensitivity across the visual field, and localised deficits (McKendrick and Badcock 2004a). It has been hypothesised that these patterns of visual field loss are underpinned by different disease mechanisms. General sensitivity is strongly correlated between left and right eyes of individuals (McKendrick and Badcock 2004a) and in some people with migraine, losses of contrast sensitivity have been found that are homonymous (Wakakura and Ichibe 1992; Sullivan-Mee and Bowman 1997; Goodwin 2011), which implies a cortical basis for the deficits (see Figure 1.8 in Chapter 1). As described in Chapter 1, there is convergent evidence for interictal cortical abnormalities in both MO and MA sufferers. It is also consistent with the cortical dysfunction revealed by the PVER abnormalities in MO and MA participants in the first experiment (Chapter 2). Thus, reduced generalised sensitivity and homonymous visual field defects may result from abnormal cortical neural processing (possibly due to hyperexcitability), or aversion and heightened discomfort during the interictal period that is common to both migraine subtypes. On the other hand, localised visual field defects and monocular and arcuate patterns of visual field defect (Lewis et al. 1989; De Natale et al. 1993; McKendrick et al. 2000; McKendrick et al. 2002; Çomoglu et al. 2003; McKendrick and Badcock 2004a; Yenice et al. 2006), indicating a pre-chiasmal origin (see Figure 1.8 of Chapter 1), have also been found in different individuals with migraine. These defects imply a different causative mechanism to the generalised and homonymous decreases in sensitivity, with several authors proposing a vasospastic origin that appears unrelated to the presence of aura (Corbett et al. 1985; Phelps and Corbett 1985; Broadway and Drance 1998; Flammer et al. 2001). Thus, in this study, it is not expected from the outset that MO and MA sufferers would differ in visual field performance.
A key challenge for experiments considering the anatomical locus for dysfunction in migraine is the fact that migraine is an episodic condition. Visual function can vary with time, both in the lead up to a migraine (Judit et al. 2000; Sand and Vingen 2000;
106 Sand et al. 2008) and post-migraine (Drummond and Anderson 1992; Sullivan-Mee and Bowman 1997; Shibata et al. 1998; McKendrick et al. 2000; McKendrick and Badcock 2004b). The intensity of interictal abnormalities in the PVER normalises just before (12 to 24 hours prior) and at the onset of an attack (Judit et al. 2000; Sand and Vingen 2000; Sand et al. 2008; Sand et al. 2009). These findings suggest that during the migaine-free interval, there is progressive dysfunction in cortical processing, which creates the ‘neurophysiological readiness’ for generating a migraine attack (Vecchia and Pietrobon 2012). Such changes in sensory processing are presumed to parallel other physiological changes involved in the build up to a migraine event, such as the normalisation of cortical excitability (Chen et al. 2009; Siniatchkin et al. 2009) and the increase in serotonin to counter an interictal deficiency (Sakai et al. 2008).
In contrast, there are case reports of atypical migraines with aura where homonymous visual field defects take time to resolve after the migraine event. Some defects disappear upon cessation of migraine symptoms some hours later (Bowerman 1989; Luu et al. 2010; Goodwin 2011), yet others can persist for much longer, i.e. days, weeks, or months (Ebner 1991; Sullivan-Mee and Bowman 1997; Razeghinejad et al. 2009). On the other hand, McKendrick and Badcock (2004b) found that localised and generalised sensitivity losses, primarily affecting one eye, were worse the day after migraine. This suggests that some visual deficits may in fact be retinal sequelae of migraine.
Given that pattern electrophysiology and visual fields both measure the integrity of visual neurones, it is possible that post-migraine changes in visual field performance are accompanied by similar changes in electrophysiology. Such a result would suggest that the changes in retinal and cortical function are sequelae of migraine, and might contribute to the heterogeneity of PVER findings in migraine by adding to inter- individual variability. Thus, for the first time, the PERG, PVER, and visual fields were measured in the same cohort of migraine sufferers on two occasions after a migraine. A secondary aim was to consider the anatomical locus of abnormalities, as
107 inferred from the spatial pattern and binocularity of visual field defects, and from comparison of simultaneously recorded PERG and PVER.
In this experiment, the following hypotheses were tested:
H2A: That measures of retinal function are worse in the days immediately after a migraine.
H2B: That measures of cortical function are worse in the days immediately after a migraine.
Parts of this chapter, including Figures 3.1, 3.4, 3.8, 3.10, 3.11, and 3.12 and Tables 3.1 to 3.6 in modified form, have been published in the following journal article, a copy of which appears in Appendix B:
Nguyen, B.N., Vingrys, A.J. and McKendrick, A.M. (2013). “The effect of duration post-migraine on visual electrophysiology and visual field performance in people with migraine.” Cephalalgia. Aug 22. [Epub ahead of print].
3.2 Methods
3.2.1 Participants
3.2.1.1 Power analysis
The main purpose of this study was to consider whether visual field and electrophysiological responses were worse in the days immediately after migraine. In the study by McKendrick and Badcock (2004b), there was a small but significant loss of generalised sensitivity the day after migraine. Concurrent with the diffuse loss,
108 there was a marked increase in the mean visual field index used to describe localised visual field deficits, which was particularly evident when testing with flickering stimuli. Hence, sample sizes for this study were determined by a power analysis based on the control group data on flicker perimetry obtained by McKendrick and Badcock (2004b). The power analysis revealed that a sample size of 17 participants per group was adequate to find an increase in localised visual field loss of three times the magnitude of normal observers on flicker perimetry using a Mann-Whitney rank sum test (80% power at α = 0.05).
Participants who volunteered for the previous experiment (Chapter 2) were invited to return for a second test session within one year of the first test session. Of the 75 participants in Experiment 1 (Chapter 2), 26 non-headache controls and 17 people with migraine (11 MO, 6 MA) returned. The MO and MA groups were pooled, as previous studies show that these migraine subgroups demonstrate similar visual field losses (Drummond and Anderson 1992; De Natale et al. 1993; McKendrick et al. 2000; McKendrick et al. 2002; Çomoglu et al. 2003; McKendrick and Badcock 2004a; McKendrick and Badcock 2004b). The control (aged 19 – 46, mean ± standard deviation: 25 ± 6 years) and migraine (aged 19 – 43, 29 ± 7 years) groups did not differ in age (Mann Whitney U = 158.5, p = 0.12).
3.2.1.2 Participant characteristics
All participants underwent a screening examination and met the eligibility criteria outlined in Chapter 2. As visual field defects can result from optic nerve head damage and retinal nerve fibre layer loss, the eye examination ensured that the participants did not show signs of glaucomatous damage on optic nerve head imaging using the Heidelberg Retinal Tomograph (HRT). There was no difference between in global rim area (main effect of group; F (2, 71) = 1.08, p = 0.34) or rim volume (F (2, 71) = 0.98, p = 0.38) of the optic nerve head, which are two HRT parameters that correlate with perimetric indices describing generalised and localised visual field loss in people with glaucoma (Iester et al. 1997).
109 The subset of migraine participants from Experiment 1 (Chapter 2) who were tested in this study were not biased to a particular migraine characteristic. With the exception of higher MIDAS questionnaire scores reported by the MO participants (Mann-Whitney U = 7.0, p = 0.01), there was no difference between MO and MA participants in their self-reported migraine characteristics, as shown in Table 3.1.
Table 3.1 – Self-reported migraine characteristics (median, minimum – maximum).
Migraine characteristic MO MA Statistic p-valuea
Days since last migraine at first visit 18 (7 – 150) 25 (7 – 200) U = 26.0 0.51
Days since last migraine at second visit 3 (1 – 5) 2 (1 – 6) U = 24.0 0.38
Age at first migraine (years) 15 (4 – 17) 12 (10 – 30) U = 21.0 0.24
Migraine history (years) 13 (3 – 23) 20 (7 – 30) t15 = 0.77 0.45
Migraines in past year 8 (1 – 50) 5 (1 – 50) U = 17.5 0.39
Weeks between migraines 3 (1 – 20) 6 (1 – 24) U = 22.0 0.29
Estimated number of lifetime attacks 100 (30 – 550) 89 (14 – 1300) U = 25.0 0.45
MIDAS questionnaire score (days) 20 (0 – 49) 3 (1 – 4) U = 7.0 0.01 **
Headache duration (hours) 12 (2 – 72) 8 (2 – 48) U = 26.0 0.51 a p-values represent the significance of group comparisons (Mann Whitney rank sum or t-tests) ** denotes significant difference between MO and MA groups (p < 0.05)
3.2.1.3 Timing of the test visit
In order to evaluate changes in visual fields and electrophysiology with time (H2A and H2B), participants were tested at two visits. Each session lasted up to three hours, including regular breaks between tests. Control participants completed two sessions at least one day apart (median 18 days, range 1 – 132 days). For people with migraine, the first visit was scheduled at least seven days after the offset of a migraine, as per the first experiment (Chapter 2). The second visit was scheduled as close as practicable, but at least one day, after the cessation of migraine symptoms (maximum six days post-migraine). Participants completed a Headache Diary (see Appendix B, Section 8.1) and/or notified by email or phone of the occurrence of
110 migraine events. The most recent migraine event was not different from the participants’ typical migraines in terms of the symptoms experienced and medications taken, as confirmed by clinical interview. The difference in the number of days post-migraine between the two visits ranged from three to 199 days (median 16 days, see Figure 3.1).
Figure 3.1 – Days since last migraine at the two test visits for migraine participants. Visit 1 was scheduled at least seven days after a migraine. Visit 2 was scheduled at a time closer after a migraine (within six days, dotted horizontal line). Data from the one participant who was tested one day before a migraine are shown as cross symbols.
Because increased PVER amplitude has been reported in the pre-attack period, up to 72 hours before a migraine (Judit et al. 2000; Sand et al. 2008; Sand et al. 2009), all participants were contacted after each test session and asked whether a migraine had occurred in the three days post-testing. This follow-up found that only one migraine participant suffered a migraine the day after the first test visit. Data from this participant have been represented as cross symbols in all figures. Excluding the data from this individual from statistical analyses did not change the conclusions of this study (see Appendix B, Section 8.4.2).
111 Prodromal symptoms, including fatigue and difficulty concentrating, commonly occur up to 48 hours before an attack (Blau 1980; Kelman 2004b), which can affect a person’s visual field performance. Eleven of the 17 migraine participants (65%) regularly experienced a prodromal phase in the few hours before the onset of their migraine headache or aura. Given that all but one participant were tested at least three days before a migraine, these prodromal symptoms were unlikely to have affected the measurements obtained in this study.
3.2.2 Pattern electrophysiology
The PERG and PVER were recorded simultaneously to assess retinal and cortical function. The methods were the same as for Experiment 1 and have been described in Chapter 2.
3.2.3 Visual field tests
Visual field tests were used to test for spatially localised deficits in contrast sensitivity. They also provided information regarding the origin of anomalies, as lesions at different anatomical locations along the visual pathway result in characteristic patterns of visual field defect (see Figure 1.8 of Chapter 1). Three different perimetric tests were chosen, all of which have identified visual field defects in people with migraine. Standard automated perimetry (SAP) was conducted as it is the standard perimetric technique and is most commonly encountered in clinical practice. Temporal modulation perimetry (TMP) and short-wavelength automated perimetry (SWAP) were also included, as visual field defects in people with migraine have been identified using TMP (McKendrick et al. 2000; McKendrick and Badcock 2004b) and SWAP (McKendrick et al. 2002) that are not measurable on SAP.
Visual field tests were performed on the same day as electrophysiological recording and were always conducted first because of possible ocular discomfort following electrode placement for PERG recordings. Participants completed SAP first, as it is
112 well tolerated and easiest for a naïve observer to learn. TMP and SWAP were conducted next, in random order. Participants had a brief practice before testing. SAP and SWAP tests were generally completed within five minutes, whereas TMP required approximately 10 minutes per test. To minimise the effect of fatigue, each examination was paused approximately every five minutes to allow a short break before continuing with the same or next test. The tests were conducted using commercially available equipment and standard clinical methods, as described below.
3.2.3.1 Equipment and room setup
Two automated perimeters were used. SAP and TMP were performed on the Medmont M-700 perimeter (Medmont Pty Ltd., Camberwell, Victoria, Australia), which has been described in detail elsewhere (Vingrys and Helfrich 1990). The Medmont perimeter consists of a hemispherical bowl that serves as the uniform background, with light-emitting diodes (LEDs) as the light sources. SWAP was performed on the Octopus 101 perimeter (Haag-Streit Inc., Koeniz, Switzerland), which uses a projection system to present the target stimuli. A detailed description of the Octopus perimeter appears in Weijland et al. (2004). The maximum field size tested by these perimeters (30°) corresponded to the extent of the checkerboard stimulus used for pattern electrophysiology (see Chapter 2). Room lights were turned off throughout testing and participants wore the appropriate refractive correction for the near working distance (Medmont: 33 cm, Octopus: 42.5 cm).
3.2.3.2 Test strategies
Standard automated perimetry
2 The achromatic stimuli for SAP (λmax = 565 nm, max luminance 320 cd/m ) subtended 0.43° in diameter (Goldmann size III). The targets were presented on a background luminance of 3.2 cd/m2 (CIE 1931 chromaticity co-ordinates: x = 0.53, y = 0.42) for 200 ms and were arranged in concentric rings around the centre of the
113 visual field (Vingrys and Helfrich 1990). In this study, SAP targets appeared at 101 locations at 1°, 3°, 6°, 10°, 15°, 22°, and 30° eccentricities. Participants were instructed to press the button when a stimulus was detected against the background luminance. SAP thresholds were measured using the Central Threshold test, which uses a maximum-likelihood thresholding algorithim (King-Smith et al. 1994), the Zippy Estimation by Sequential Testing (ZEST) procedure (Turpin et al. 2002). Estimates of SAP sensitivity were given in 1 dB steps.
Temporal modulation perimetry
The stimuli for TMP were identical to those used for SAP (λmax = 565 nm, 0.43° diameter or Goldmann size III), except for the 800 ms stimulus duration, and were presented at 73 locations at 1°, 3°, 6°, 10°, 15° and 22° eccentricities. The TMP stimuli consisted of a flickering component that was superimposed on a pedestal of steady light (i.e. luminance pedestal flicker). TMP therefore measures the amount of flicker that is needed to distinguish the flicker from a steady, uniform stimulus. Participants were instructed to respond if the target detected was perceived as ‘flickering’, ‘twinkling’, ‘shimmering’, etc and not to respond when the stimulus was perceived as static. The temporal frequency of the flickering stimuli was automatically varied with retinal eccentricity using the Auto-Flicker test on the Medmont perimeter, with stimuli at more peripheral locations flickering at a slower rate (18 Hz at 1° – 3°, 16 Hz at 6°, 12 Hz at 10° – 15°, and 9 Hz at 22°). Detection thresholds were measured with a 6-3 staircase procedure, as described in more detail by Vingrys and Helfrich (1990). Estimates of sensitivity were therefore given in 3 dB steps.
Short-wavelength automated perimetry
For SWAP, blue (λmax = 440nm) test stimuli of 1.72° diameter (Goldmann size V) were projected for 200 ms against a yellow background (100 cd/m2) at 52 locations at 3°, 9°, 15°, and 21° eccentricities. Participants were instructed to press the button
114 whenever a blue target was detected against the background luminance. Adaptation to the yellow background was required before commencing each test. Most studies using blue-on-yellow perimetry in the literature report adaptation times of between 2 and 5 minutes (Wild et al. 1998; Eisner et al. 2004; Bengtsson and Heijl 2006; Anderson et al. 2009). In this study, an adaptation time of three minutes was adopted. Detection thresholds were measured using the ‘Dynamic’ strategy (Weber and Klimaschka 1995), which is a staircase procedure that varies the step size depending on the depth of the defect, from 2 dB in regions of normal sensitivity to 10 dB where sensitivity is depressed. Estimates of SWAP sensitivity were given in 1 dB steps.
3.2.3.3 Reliability indices
Three reliability indices were automatically generated by the perimeters, described below.
False positives
Up to 10% of the total number of presentations were ‘catch trials’ (Vingrys and Demirel 1998) that were used to monitor the number of false-positives during testing. For SAP and SWAP, a false-positive occurred when a person responded to a non-existent stimulus. In the case of TMP, static stimuli were presented as false- positive catch trials. Visual field results with false-positive rates exceeding 30% were discarded (Anderson 1992), as this would produce artificially elevated sensitivity (Demirel and Vingrys 1994).
False negatives
A false-negative indicated a failure to respond to a stimulus at a visual field location that was previously ‘seen’ at a lower brightness level. This result could occur if the participant was unreliable. Alternatively, high false-negative rates might also occur
115 with flattening of the psychometric function, or frequency-of-seeing curve, in areas of reduced sensitivity, such that the test-retest variability of single threshold estimates can encompass the dynamic range of the perimeter (Artes et al. 2002). By convention, visual field results with false-negative rates exceeding 30% were excluded from analysis (Anderson 1992; Medmont M700 Automated Perimeter User Manual 2007).
Fixation losses
During visual field testing, participants were required to maintain central fixation. Fixation losses were identified by the Medmont perimeter using an automated blind- spot monitor (Heijl and Krakau 1977), which periodically presented a stimulus of maximum intensity at the expected position of the blind-spot based on normal population data for optic nerve head location, i.e. approximately 15° temporal of the fovea (Vingrys and Helfrich 1990). Responses to the blind-spot stimuli were counted as fixation losses.
The typical acceptable criterion for fixation loss is 30% (Anderson 1992). In this study, the automated blind-spot monitor of the Medmont perimeter identified fixation losses exceeding 30% in four control and four migraine participants. This can occur due to a limitation of the Heijl-Krakau method (Heijl and Krakau 1977) in accurately mapping the blindspot. In theory, the stimuli should yield no response if the patient is maintaining steady fixation. However, some people have an anatomical position of the blind-spot that differs from the population average, which can result in a false measurement of ‘100% fixation loss’ even when the observer is clearly making no eye movements (Sanabria et al. 1991). Thus, to supplement the automated blind-spot monitor, steady fixation was confirmed by monitoring the limbal position of the observer’s eye during testing by direct visual inspection (Medmont perimeter) or via an in-built infra-red video camera (Octopus perimeter). Eye movements larger than 2 to 3° that affect visual field outcomes (Demirel and Vingrys 1994) have been shown to be detectable by these methods.
116 3.2.4 Visual field analysis
3.2.4.1 Global indices used to describe visual field loss
To characterise generalised and localised visual field loss, the perimeters automatically generated global indices relative to the proprietary age-matched internal normative databases. These were (1) Average or Mean Defect and (2) Pattern Defect or Loss Variance, respectively.
(1) The Average or Mean Defect indicates whether there is a generalised depression or elevation in visual field sensitivity. The index is calculated using Equation 3.1 by comparing an individual’s performance at each visual field location to a database of age-matched norms:
1 m ∑(zi − x i ) (Equation 3.1) m i=1
where m is the number of test locations (not including the blind-spot), zi is the age- € corrected normal sensitivity value at test location i, and x i is the person’s mean sensitivity at location i. For the Medmont perimeter, a negative Average Defect € € represents a visual field that is, on average, worse than expected for a person of the € € same age (Medmont M700 Automated Perimeter User Manual 2007). In contrast, € the higher the Mean Defect on the Octopus perimeter, the worse the generalised sensitivity (Weijland et al. 2004).
(2) The Pattern Defect or Loss Variance is an index designed to emphasise localised losses, or asymmetry in the visual field. It is determined by comparing the performance of each individual to the sensitivity expected based on the individual’s own hill-of-vision, on a point-by-point basis. Pattern Defect and Loss Variance are always positive values. A large value indicates a local depression relative to the
117 remainder of the visual field, whereas if deviations from the person’s hill-of-vision are distributed randomly throughout the visual field, then the index will be small.
Currently, full mathematical details of the calculation of Pattern Defect on the Medmont perimeter are not available from the manufacturer. On the other hand, Loss Variance on the Octopus perimeter is the squared value of the variance of the local defects and is calculated using Equation 3.2:
m 1 2 ∑(zi − x i − MD) (Equation 3.2) m −1 i=1
where m is the number of test locations (not including the blind-spot), zi is the age- € corrected normal sensitivity value at test location i, x i is the estimated mean sensitivity over all repeats at location i, and MD is the Mean Defect of the Octopus € € perimeter, as calculated using Equation 3.1. € € € € 3.2.4.2 Point-wise comparisons
A limitation of using global indices alone for comparisons between control and migraine groups is that people with migraine are not excluded from the proprietary databases from which the indices are derived. In addition, global indices do not illustrate which locations are abnormal across the visual field. Hence, point-wise comparisons were used to evaluate an individual’s response against the performance of the 26 control participants at each visual field location.
This method of point-wise comparisons has been described in detail previously (McKendrick 1999; McKendrick et al. 2000). Briefly, locations at and immediately above and below the blindspot were not included in the analysis. As visual field sensitivity is non-Gaussian (Vingrys and Pianta 1998), the empirical 8th and 92nd percentiles of the control group were determined at every location, which corresponded to the 2nd best and 2nd worst performing control participant. Locations
118 where sensitivity was lower than the 8th percentile were flagged as ‘depressed’, or ‘better’ if sensitivity was above the 92nd percentile limit (at p < 0.08). Point-wise confidence limits were also determined for the change in sensitivity between the two test visits, where a negative change indicated a reduction in sensitivity at the second visit.
The number of statistically abnormal points required to flag the overall visual field result as abnormal was subsequently determined, assuming that each point behaved independently. The two-tailed probability that an individual point fell outside the confidence interval range yet still belonged to the normal sample was α = 0.04. The cumulative probability that a visual field contained at least n number of statistically abnormal points (p < 0.04 for a single location) depended on α and the total € number of test locations ( N ), and was given by: €
€ N C n (1 )N−n (Equation 3.3) n ⋅α ⋅ −α€
where €
N! N C = (Equation 3.4) n n!(N − n)!
The number of test locations varied depending on the perimetric task (101 test € points on SAP, 73 on TMP, and 50 on SWAP). The cumulative probability of identifying eight locations on SAP, six locations on TMP, and five locations on SWAP that were abnormal out of all the locations tested were 0.030, 0.045, and 0.035, respectively (see Appendix B, Section 8.2.1). Accordingly, visual fields were deemed abnormal overall (p < 0.05) if there were at least eight locations on SAP, six locations on TMP, or five locations on SWAP that were identified as abnormal (α = 0.04). These probabilities were based on an assumption of independence between neighbouring points. While this assumption may not be strictly valid, it represents a conservative criterion.
119 3.2.4.3 Characterisation of the pattern of visual field loss
Once locations with abnormal sensitivity were identified using the point-wise analysis described above, and given a significant number of statistically abnormal points in either eye, visual fields of both eyes were examined to classify whether the pattern of defect was monocular or bilateral. As illustrated by Figure 1.7 in Chapter 1, lesions at particular locations along the visual pathway result in characteristic patterns of visual field loss. Monocular visual field defects result from lesions prior to the optic chiasm (e.g. at the retina). Binocular and homonymous defects are a feature of post-chiasmal dysfunction (e.g. at the brain), although the presence of a bilateral defect does not preclude the possibility of two monocular defects.
For binocular defects, two approaches were used to determine if the defect was homonymous: (1) visual inspection, to see if locations of depressed sensitivity respected the vertical midline; and (2) quadrant analysis (McKendrick and Badcock 2004c), where a quadrant was classified as abnormal (p < 0.05) if there were at least four locations on SAP, three locations on TMP, or three locations on SWAP that were identified as abnormal (for cumulative probabilities of each quadrant, see Appendix B, Section 8.2.1). When the same quadrant was classified as abnormal in both eyes, using either criterion, the deficit was considered homonymous.
3.2.5 Statistical analysis
3.2.5.1 Worst eye analysis
For control and migraine participants, a ‘worst eye’ was chosen for analysis based on the total number of abnormal points across all visual fields (see Appendix B, Section 8.2.2), and is represented in all figures and tables of this chapter. One migraine participant demonstrated a false negative rate greater than 30% in the left eye on one of the visual field tests. In this case, the right eye was considered the ‘worst eye’ and only data from the right eye were included in the subsequent analyses. If the number
120 of abnormal points was the same in the two eyes, one eye was chosen at random. The proportion of right eyes that were designated as the worst eye did not differ between control and migraine groups (controls 50%, migraine 53%, chi-square: p = 0.85).
3.2.5.2 Group comparisons
Group comparisons in SPSS version 20.0 (SPSS Inc., Chicago, IL, USA) were performed using a repeated-measures analysis of variance (RM-ANOVA) with a between-factor of group (control versus migraine) and a within-factor of visit (first versus second visit). The interaction effect was tested to address the two hypothesis of this study (H2A and H2B), as a significant interaction between group and visit would indicate that the difference between the two groups depended on the time of testing. Paired independent sample t-tests or paired Wilcoxon signed rank tests were subsequently conducted to determine whether visual responses were significantly changed compared to baseline on an individual basis.
3.2.5.3 Individual comparisons
To determine whether an individual with migraine showed worse performance in the days immediately after a migraine, differences in performance between visits were calculated such that a negative difference signified worse results at the second visit, and vice versa. Individual migraine differences were plotted against the range of control test-retest variability, the boundaries of which were nominally given by the 2nd worst (8th percentile) and 2nd best (92nd percentile) performing control participant. These ‘discrepancy’ plots (Figures 3.3, 3.5, 3.7, and 3.9) aim to assist in a qualitative interpretation of the data by identifying individual departures from normal test-retest variability. Given a sample size of 17 migraine participants, one person with migraine was expected to fall outside the normal test-retest limits.
121 3.3 Results
3.3.1 Changes in the electrophysiological response
3.3.1.1 The pattern electroretinogram (retinal response)
Figure 3.2 depicts the group mean and individual PERG amplitude data. The corresponding discrepancy plots are shown in Figure 3.3. In the previous experiment (Chapter 2), the PERG was not different between control and migraine groups in terms of amplitude. This result was again observed in this study, where PERG amplitudes were similar at both visits (main effect of group: F (1, 41) = 0.73, p = 0.40). The raw waveforms and corresponding Fourier spectra of the 25th, 50th, and 75th percentiles of change in PERG amplitude from one visit to the next are depicted in Appendix B, Section 8.3. A novel addition to the literature was the consideration of changes in the PERG after migraine. As it is possible that any of the PERG parameters could change with time post-migraine, the interactions between group and visit were considered for the various PERG components separately. These interactions did not reach significance at p < 0.05 and are outlined in Table 3.2.
Likewise, PERG timing parameters were, overall, not different between control and migraine groups (main effect of group: F (1, 41) = 0.51, p = 0.48), and were not significantly changed with time post-migraine as shown in Table 3.2. Taken together, these results imply that global retinal function, as assayed by the PERG, is normal in between migraines, even when tested in the days immediately after migraine.
122
Figure 3.2 – Group mean and individual PERG amplitudes at the two test visits. (A) Transient PERG P50 amplitude (controls: n = 26, migraine: n = 17) (B) Transient PERG N95 amplitude (controls: n = 26, migraine: n = 17) (C) Steady-state PERG 2F amplitude (controls: n = 26, migraine: n = 17). For migraine participants, visit 1 was scheduled at least seven days after a migraine, whereas visit 2 was scheduled at a time closer after a migraine (within six days). Data from the one participant who was tested one day before a migraine are shown as cross symbols. Error bars represent 95% confidence intervals of the mean.
123
Figure 3.3 – Discrepancy plots comparing PERG amplitude at the two test visits of the migraine individuals (n = 17). (A) Transient PERG P50 (B) Transient PERG N95 (C) Steady-state PERG 2F. Data from the one participant who was tested one day before a migraine are shown as cross symbols. Dashed horizontal lines represent the 2nd worst (8th percentile) and 2nd best (92nd percentile) performing participant of the control group (n = 26). A negative difference (y-axis) denotes reduced amplitude at the second visit, as shown in the bottom half of each panel. The shaded area indicates when individual migraine performance was worse than the 2nd worst control participant in the days immediately after an attack. The mean bias and 95% confidence interval of the differences is shown to the right of each panel.
124 Table 3.2 – Summary of PERG parameters (mean ± standard deviation).
Controls Migraine
Parameter Visit 1 Visit 2 Visit 1 Visit 2 p-valuea
Transient P50 amplitude (µV) 6.5 ± 1.4 6.5 ± 1.3 7.1 ± 2.2 7.0 ± 2.0 0.23
N95 amplitude (µV) 10.2 ± 2.6 9.8 ± 2.1 11.6 ± 3.2 10.1 ± 2.2 0.25
N35 time (ms) 22 ± 3 22 ± 3 21 ± 2 22 ± 3 0.44
P50 time (ms) 51 ± 3 51 ± 2 51 ± 2 51 ± 2 0.65
N95 time (ms) 104 ± 5 106 ± 4 103 ± 3 106 ± 4 0.94
Steady-state 2F amplitude (µV) 3.3 ± 0.6 3.3 ± 0.7 3.2 ± 1.0 3.3 ± 0.9 0.85
2F phase (π rads) 2.1 ± 0.1 2.1 ± 0.1 2.1 ± 0.1 2.1 ± 0.1 0.38
PERG ratio 1.2 ± 0.2 1.2 ± 0.2 1.3 ± 0.3 1.3 ± 0.3 0.70 a p-values represent the RM-ANOVA interaction between group (control, migraine) and visit (visit 1, visit 2)
3.3.1.2 The pattern visual evoked response (cortical response)
The PVER amplitude data are illustrated in Figure 3.4, with a depiction of the raw waveforms appearing in Appendix B, Section 8.3. In Chapter 2, deficits in PVER amplitude were found in the transient and steady-state responses of people with migraine, which were dependent on the component. A significant interaction between component and group (F (2, 80) = 13.55, p < 0.001) was again observed in this study. Hence, the different components were analysed separately to determine if any of these measures changed significantly with time post-migraine (Table 3.3).
For the transient P100 amplitude (Figure 3.4A), the interaction between group and visit was significant (F (1, 41) = 5.23, p = 0.03). The difference between migraine and control groups appeared to be more apparent at the second visit, with a slight decrease in control group amplitudes and a slight increase in migraine group amplitudes (Figure 3.4A). However, paired t-tests showed that neither group demonstrated a significant change in P100 amplitude from baseline (controls t25 =
1.63, p = 0.12; migraine t16 = 1.58, p = 0.13) and none of the individuals with migraine showed a reduction in amplitude that was greater than the 2nd worst
125 performing control participant (Figure 3.5A). Thus, the data were not supportive of a significant change in P100 amplitude in the people with migraine that exceeded normal test-retest variability. Previous reports demonstrate an increase in the transient N135 amplitude within the 72 hours before an attack, whereas no changes were evident after an attack (Sand et al. 2008; Sand et al. 2009). In this study, despite a trend for an interaction between group and visit (Figure 3.4B; F (1, 41) = 3.62, p = 0.06), paired t-tests did not find a significant change in N135 amplitude from one visit to another in the migraine group (t16 = 0.88, p = 0.39).
The steady-state PVER amplitude was reduced at both visits (Figure 3.4C; main effect of group: F (1, 40) = 11.43, p = 0.002). This confirmed that the previous findings of steady-state PVER deficits in aura and non-aura groups in Experiment 1 (Chapter 2) were repeatable. However, changes in the steady-state PVER amplitude with time post-migraine were not evident (group x visit interaction: F (1, 40) = 0.57, p = 0.45). Nevertheless, the corresponding discrepancy plot (Figure 3.5C) shows a bias in the migraine group for greater deficits in steady-state PVER amplitude at the second visit, within six days of a migraine. There is an uneven distribution of individual migraine test-retest differences, which are mostly concentrated in the bottom half of the Figure 3.5C, indicating worse performance at the second visit.
This trend is also seen in the paired t-tests between the first and second visits (t15 = 1.88, p = 0.08). Figure 3.5C also shows, however, that only one individual with migraine fell below the lower limit of test-retest variability defined by the 2nd worst performing control participant.
PVER timing in people with migraine was consistently within normal limits at both visits (main effect of group: F (1, 40) = 0.16, p = 0.70) and did not become worse closer to a migraine (Table 3.3; group x visit interactions: p > 0.05). Since the timing and amplitude of the PERG and PVER remained the same at both visits, there was no change in the retinocortical time with time post-migraine (Table 3.3; group x visit interactions: p > 0.05).
126
Figure 3.4 – Group mean and individual PVER amplitudes at the two test visits. (A) Transient PVER P100 amplitude (controls: n = 26, migraine: n = 17) (B) Transient PVER N135 amplitude (controls: n = 26, migraine: n = 17) (C) Steady-state PVER 2F amplitude (controls: n = 26, migraine: n = 16). For migraine participants, visit 1 was scheduled at least seven days after a migraine, whereas visit 2 was scheduled at a time closer after a migraine (within six days). Data from the one participant who was tested one day before a migraine are shown as cross symbols. Error bars represent 95% confidence intervals of the mean.
127
Figure 3.5 – Discrepancy plots comparing PVER amplitude at the two test visits of the migraine individuals (n = 17). (A) Transient PVER P100 (B) Transient PVER N135 (C) Steady-state PVER 2F. Data from the one participant who was tested one day before a migraine are shown as cross symbols. Dashed horizontal lines represent the 2nd worst (8th percentile) and 2nd best (92nd percentile) performing participant of the control group (n = 26). A negative difference (y-axis) denotes reduced amplitude at the second visit, as shown in the bottom half of each panel. The shaded area indicates when individual migraine performance was worse than the 2nd worst control participant in the days immediately after an attack. The mean bias and 95% confidence interval of the differences is shown to the right of each panel.
128 PVER interhemispheric asymmetry has been associated with migraine symptoms – specifically laterality of headache (Logi et al. 2001) and aura (Tagliati et al. 1995; Shibata et al. 1998; Khalil et al. 2011). However, there was no change in PVER asymmetry with time post-migraine (Table 3.4, paired Wilcoxon signed rank tests: p > 0.05). In summary, the PVER amplitude deficits found in Experiment 1 (Chapter 2) at least seven days post-migraine were again present but unchanged in the days immediately after a migraine.
Table 3.3 – Summary of PVER parameters (mean ± standard deviation).
Controls Migraine
Parameter Visit 1 Visit 2 Visit 1 Visit 2 p-valuea
Transient P100 amplitude (µV) 11.3 ± 5.0 10.6 ± 5.3 12.9 ± 4.8 13.8 ± 5.1 0.03
N135 amplitude (µV) 10.0 ± 2.5 9.8 ± 2.1 11.3 ± 3.2 10.9 ± 2.8 0.06
N75 time (ms) 72 ± 7 71 ± 8 71 ± 4 69 ± 7 0.62
P100 time (ms) 103 ± 6 102 ± 6 103 ± 6 102 ± 4 0.32
N135 time (ms) 146 ± 15 144 ± 15 143 ± 8 146 ± 8 0.09
Retinocortical time (ms) 51 ± 5 51 ± 6 51 ± 4 51 ± 5 0.44
PVER/PERG ratio 1.2 ± 0.5 1.1 ± 0.5 1.2 ± 0.6 1.2 ± 0.5 0.30
Steady-state 2F amplitude (µV) 3.9 ± 1.9 3.7 ± 2.0 2.2 ± 1.4 ** 1.9 ± 1.0 ** 0.45
2F phase (π rads) 1.5 ± 0.4 1.5 ± 0.3 1.5 ± 0.2 1.6 ± 0.4 0.82
PVER/PERG ratio 1.3 ± 0.6 1.2 ± 0.7 0.7 ± 0.5 ** 0.6 ± 0.5 ** 0.13 a p-values represent the RM-ANOVA interaction between group (control, migraine) and visit (visit 1, visit 2) ** denotes significant reduction in steady-state PVER amplitude in migraine group at both visits [RM-ANOVA main effect of group: F (1, 40) = 11.43, p = 0.002]
Table 3.4 – PVER interhemispheric asymmetry (median, minimum – maximum).
Controls Migraine
Amplitude a a Visit 1 Visit 2 p-value Visit 1 Visit 2 p-value asymmetry (%)
Transient P100 18 (1 – 42) 21 (3 – 62) 0.24 16 (2 – 56) 16 (2 – 43) 0.69
Transient N135 22 (3 – 49) 12 (2 – 37) 0.10 15 (2 – 70) 16 (2 – 42) 0.49
Steady-state 2F 15 (2 – 41) 20 (3 – 51) 0.29 20 (3 – 60) 29 (3 – 75) 0.38 a p-values represent the effect of visit (paired Wilcoxon signed rank tests)
129 3.3.2 Changes in visual field sensitivity
3.3.2.1 Global indices of generalised visual field loss
On the Medmont perimeter, a generalised depression in sensitivity is indicated by a lower Average Defect value. In contrast, increased Mean Defect values on the Octopus perimeter are associated with greater sensitivity loss. Hence, although raw Mean Defect values are depicted in Figures 3.6C and 3.7C, the inverse of Mean Defect was included in the RM-ANOVA, such that lower Mean Defect indicated worse visual field results. This ensured that effects were in the same direction across the different perimetric tasks. The RM-ANOVA revealed that Average Defect and Mean Defect on SAP, TMP, and SWAP were similar between groups (main effect of group: F (1, 41) = 0.98, p = 0.33). However, generalised sensitivity became worse at the visit closer to a migraine (group x visit interaction: F (1, 41) = 5.44, p = 0.03).
Additional RM-ANOVAs were run to investigate each perimetric task individually (Table 3.5). For SAP Average Defect, there was a significant interaction between group and visit (Figure 3.6A; interaction: F (1, 41) = 5.17, p = 0.03). The discrepancy plot (Figure 3.7A) indicates a bias towards improved SAP Average Defect at the second visit, where test-retest differences were mostly positive. This improvement in generalised sensitivity was evident in the control group only (paired t-test: t25 = 4.80, p < 0.001). The migraine participants did not show the same learning benefits as controls (paired t-test: t16 = 1.08, p = 0.30). Figures 3.6B and 3.6C demonstrate a decrease in mean TMP Average Defect and an increase in SWAP Mean Defect, respectively, which both indicate worse generalised sensitivity at the second visit. Although these changes did not reach statistical significance (group x visit interaction: TMP F (1, 41) = 0.54, p = 0.47; SWAP F (1, 41) = 3.14, p = 0.08), the discrepancy plots (Figures 3.7B and 3.7C) identify more than individual with migraine with a greater reduction in generalised sensitivity than expected from controls (p = 0.04).
130
Figure 3.6 – Group mean and individual global indices of generalised sensitivity at the two test visits. (A) SAP Average Defect (controls: n = 26, migraine: n = 17) (B) TMP Average Defect (controls: n = 26, migraine: n = 17) (C) SWAP Mean Defect (controls: n = 26, migraine: n = 17). For migraine participants, visit 1 was scheduled at least seven days after a migraine, whereas visit 2 was scheduled at a time closer after a migraine (within six days). Data from the one participant who was tested one day before a migraine are shown as cross symbols. Error bars represent 95% confidence intervals of the mean. Note that lower Average Defect values (Panels A and B) indicate a generalised depression, whereas lower Mean Defect values (Panel C) reflect higher sensitivity.
131
Figure 3.7 – Discrepancy plots comparing Average/Mean Defect at the two test visits of the migraine individuals (n = 17). (A) SAP Average Defect (B) TMP Average Defect (C) SWAP Mean Defect. Data from the one participant who was tested one day before a migraine are shown as cross symbols. Dashed horizontal lines represent the 2nd worst (8th percentile) and 2nd best (92nd percentile) performing participant of the control group (n = 26). A negative difference (y-axis) denotes reduced generalised sensitivity at the second visit, as shown in the bottom half of each panel. The shaded area indicates when individual migraine performance was worse than the 2nd worst control participant in the days immediately after an attack. The mean bias and 95% confidence interval of the differences is shown to the right of each panel.
132 Table 3.5 – Average Defect and Mean Defect (mean ± standard deviation).
Controls Migraine
Test Visit 1 Visit 2 Visit 1 Visit 2 p-valuea
SAP Average Defect -0.9 ± 1.1 0 ± 1.3 -0.9 ± 1.0 -0.7 ± 1.1 0.03
TMP Average Defect -3.2 ± 1.0 -3.2 ± 1.1 -3.4 ± 1.2 -3.6 ± 1.2 0.47
SWAP Mean Defect 3.1 ± 2.9 2.9 ± 3.0 3.3 ± 2.9 3.9 ± 2.7 0.08 a p-values represent the RM-ANOVA interaction between group (control, migraine) and visit (visit 1, visit 2)
3.3.2.2 Global indices of localised visual field loss
The global indices of localised sensitivity loss, Pattern Defect and Loss Variance, are illustrated in Figure 3.8. These data were not normally distributed. Therefore, paired Wilcoxon signed rank tests were used to evaluate the change in Pattern Defect and Loss Variance with time post-migraine. These comparisons were not significant (Table 3.6). However, the corresponding discrepancy plots (Figure 3.9) show Pattern Defect and Loss Variance data that fall below the 8th percentile of the control group (Figure 3.9, lower dashed horizontal lines), implying that localised sensitivity loss was worse in some migraine individuals in the days immediately after an attack.
Because there was no difference between the first and second visits in either group, Pattern Defect and Loss Variance were averaged across the two visits to enable overall comparisons between control and migraine groups. For SAP and TMP, the difference in average Pattern Defect between groups did not reach statistical significance (Mann Whitney rank sum test: SAP U = 163.0, p = 0.15; TMP U = 159.0, p = 0.13). The average SWAP Loss Variance was significantly higher in the migraine group (Mann Whitney rank sum test: U = 129.5, p = 0.02), indicating localised sensitivity loss at both visits.
133
Figure 3.8 – Group median and individual global indices of localised visual field loss at the two test visits. (A) SAP Pattern Defect (controls: n = 26, migraine: n = 17) (B) TMP Pattern Defect (controls: n = 26, migraine: n = 17) (C) SWAP Loss Variance (controls: n = 26, migraine: n = 17). For migraine participants, visit 1 was scheduled at least seven days after a migraine, whereas visit 2 was scheduled at a time closer after a migraine (within six days). Data from the one participant who was tested one day before a migraine are shown as cross symbols. Error bars represent the interquartile range. Higher indices indicate greater localised visual field loss.
134
Figure 3.9 – Discrepancy plots comparing Pattern Defect/Loss Variance at the two test visits of the migraine individuals (n = 17). (A) SAP Pattern Defect (B) TMP Pattern Defect (C) SWAP Loss Variance. Data from the one participant who was tested one day before a migraine are shown as cross symbols. Dashed horizontal lines represent the 2nd worst (8th percentile) and 2nd best (92nd percentile) performing participant of the control group (n = 26). A negative difference (y-axis) denotes increased localised sensitivity loss at the second visit, as shown in the bottom half of each panel. The shaded area indicates when individual migraine performance was worse than the 2nd worst control participant in the days immediately after an attack. The mean bias and 95% confidence interval of the differences is shown to the right of each panel.
135 Table 3.6 – Pattern Defect and Loss Variance (median, minimum – maximum).
Controls Migraine
Test Visit 1 Visit 2 p-valuea Visit 1 Visit 2 p-valuea
1.4 1.7 3.1 2.0 SAP 0.16 0.16 (0 – 4.6) (0 – 5.5) (0 – 20.4) (0 – 20.3)
2.30 1.6 2.3 4.3 TMP 0.12 0.96 (0 – 8.1) (0 – 9.6) (0 – 18.3) (0 – 16.3)
7.8 6.4 13.6 11.2 SWAP 0.36 0.80 (2.2 – 33.8) (2.0 – 20.5) (3.5 – 113.9) (3.7 – 115.5) a p-values represent the effect of visit (paired Wilcoxon signed rank tests)
3.3.2.3 Point-wise comparisons identifying locations of visual field loss
When visual fields of migraine individuals were assessed using point-wise analysis, the majority of locations that were identified as ‘abnormal’ were depressed, not better (Figure 3.10, black bars). This was evident for all visual field tasks, although different people were identified as abnormal for each test. The total number of depressed points for a given individual did not change with time post-migraine (paired Wilcoxon signed rank tests: SAP Z = -0.05, p = 0.96; TMP Z = -0.89, p = 0.38; SWAP Z = -0.32, p = 0.75). However, the migraine group showed point-wise changes in sensitivity that fell outside that predicted from control group test-retest variability, as evidenced by the significantly greater number of points where sensitivity was further reduced at the second visit (main effect of group: F (1, 41) = 35.5, p < 0.001). The proportion of migraine participants with at least eight points that became worse after migraine on SAP, six on TMP, and five on SWAP was 41% for SAP, 24% for TMP, and 47% for SWAP. These proportions were significantly different from controls (chi-square test of proportions: SAP p = 0.008, TMP p = 0.049, SWAP p = 0.003). In contrast, the total number of abnormal points that significantly improved at the second visit was not different between groups (main effect of group: F (1, 41) = 2.83, p = 0.10). The proportion of control versus migraine participants who showed a significant number of improved points was not different between groups (SAP p = 0.24, TMP p = 0.14, SWAP p = 0.05).
136
Figure 3.10 – Proportion of the total number of visual field locations identified as abnormal based on point-wise comparisons to control sensitivity. (A) SAP (B) TMP (C) SWAP. Locations that were ‘depressed’ (lower sensitivity than 8th percentile limit) are indicated by black bars, and locations that were ‘better’ (higher sensitivity than 92nd percentile limit) are shown as white bars. A visual field was considered abnormal overall if there were at least eight SAP, six TMP, or five SWAP locations (horizontal dotted lines) that were identified as depressed. Participants 12 to 17 are the migraine with aura group (shaded area).
137 To illustrate the point-wise changes, Figure 3.11 shows the sensitivity at the first visit as a function of sensitivity at the second visit. Confidence limits were determined for the range of sensitivity values, pooled across all visual field locations. Only sensitivity values appearing at least 20 times in the dataset were included in the analysis in order to obtain estimates of the 5th and 95th confidence limits. Consistent with previous literature (Artes et al. 2002; McKendrick and Badcock 2004b), there was increased variability for locations with low sensitivity. Whereas on average the control and migraine groups showed similar upper limits of test-retest performance (Figure 3.11, filled squares), the lower limits of the migraine group were below that of controls across most of the sensitivity range (Figure 3.11, filled circles). Thus, people with migraine showed a significant number of points with reduced sensitivity to begin with (Figure 3.10) and which were associated with larger losses in the days after a migraine event (Figure 3.11).
3.3.3 Patterns of visual field loss
Point-wise comparisons not only provided information about the number of visual field locations with decreased sensitivity, but also the location of abnormality. Where point-wise comparisons revealed a statistically significant number of depressed points in the worst eye (i.e. at least eight locations on SAP, six on TMP, and five on SWAP), the fellow eye was examined to determine whether the pattern of visual field loss involved one or both eyes. The fellow eye could not be analysed for the single migraine participant with a false negative rate exceeding 30%, as only data from the one eye (designated as the ‘worst eye’) were included. Hence, 16 migraine participants were included in this analysis of the pattern of defects.
138
Figure 3.11 – Test-retest variability in visual field performance. Visual field sensitivity at the second visit as a function of sensitivity at the first visit. (A) SAP (B) TMP (C) SWAP. The shaded area is the 90% confidence interval of test-retest sensitivity for the control group. The 5th (circles) and 95th (squares) percentile limits for the migraine group are shown as individual symbols. For (A) SAP and (C) SWAP, sensitivity was measured in 1 dB steps, whereas for (B) TMP, sensitivity was measured in 3 dB steps.
139 Three people with migraine (19%) showed normal results for every visual field task at every visit, compared with 20 non-headache control participants (77%). The remainder of participants showed monocular and bilateral visual field defects. Five migraine participants (31%) demonstrated a repeatable bilateral visual field defect, an example of which is shown in Figure 3.12A, where visual field loss was diffuse across both eyes at both visits. None gave a homonymous pattern respecting the vertical midline, but the majority of cases (80%) satisfied our less conservative definition for homonymous deficits, where the same quadrant was flagged as ‘abnormal’ in both eyes. However, the quadrants flagged were not confined to one hemifield only, as the sensitivity loss was diffuse and affected the entire field.
On the other hand, four different migraine participants (25%) showed monocular sensitivity loss affecting the same eye at both visits. A further three people (19%) showed normal fields at the first visit, but developed a monocular field defect closer to a migraine. An example is illustrated in Figure 3.12B, showing inferior arcuate loss in the right eye only, one day after migraine, which was not present at the first visit 56 days after a migraine. Monocular defects ranged from patchy loss affecting all four quadrants of a single eye, to an arcuate scotoma that crossed the vertical midline.
140
Figure 3.12 – Example SAP visual field defects in people with migraine. (A) Binocular visual field defect at both visits in a 32-year-old female with migraine without aura (B) Monocular visual field defect one day after migraine in a 36-year-old female with migraine with aura. Shaded squares indicate ‘depressed’ locations where sensitivity fell below the 8th percentile of control sensitivity. A SAP visual field was considered abnormal overall (p < 0.05) if there were at least eight locations identified as depressed, relative to control group performance (α = 0.04).
141 3.3.4 Relationship between visual function and migraine features
Of the visual field and electrophysiological measures analysed in this study, the steady-state PVER amplitude (Figure 3.4C), Pattern Defect/Loss Variance (Figure 3.8), and the number of depressed points based on point-wise comparisons (Figure 3.10) remained consistently abnormal during the post-migraine period. Spearman rank correlations between these measures, averaged across both visits, were not significant (Table 3.7).
If greater migraine severity is associated with worse visual responses after an attack, then people with longer headache duration even with analgesic treatment or higher MIDAS questionnaire scores might show worse visual performance at the second visit. Indeed, McKendrick and Badcock (2004a) found a weak but statistically significant correlation between generalised visual field sensitivity to flickering stimuli and increased migraine frequency. However, in this study, correlation coefficients determined between the visual functional measures that were consistently abnormal noted above and the migraine characteristics shown in Table 3.1 were not significant (p > 0.05, see Appendix B, Section 8.4.3).
Table 3.7 – Relationship between abnormal electrophysiological measures and visual field indices averaged across both visits of the migraine group (n = 17).
a Relationship with steady-state PVER amplitude R p-value
SAP Pattern Defect -0.33 0.19
Depressed points -0.18 0.48
TMP Pattern Defect 0.07 0.80
Depressed points 0.05 0.84
SWAP Loss Variance -0.28 0.27
Depressed points -0.19 0.45 a p-values represent the significance of Spearman rank correlations
142 3.4 Discussion
The experiment described in this chapter was designed to test the following hypotheses:
H2A: That measures of retinal function are worse in the days immediately after a migraine.
This study included visual assessment at two visits: at least seven days after a migraine, and within six days of a migraine. To address the first hypothesis (H2A), the retinal electrophysiological response (PERG) was measured on the same days as visual field testing. People with migraine showed more individual locations in the visual field with further reductions in sensitivity at the second visit, one to six days after a migraine (p < 0.001). On average, sensitivity at a given location was lower than controls across the range of sensitivities observed (Figure 3.11). In some migraine participants, the loss of sensitivity was in one eye only, implying a retinal origin for the visual field deficit (e.g. Figure 3.12B). Despite the localised and monocular visual field changes in the days immediately after migraine, PERG parameters were normal in the migraine group at both visits with no further deficit closer to a migraine (Table 3.2; main effect of group: p > 0.05, group x visit interactions: p > 0.05). Thus, the PERG data do not provide support for the first hypothesis (H2A).
H2B: That measures of cortical function are worse in the days immediately after a migraine.
The PVER was also recorded on the same day as visual field testing, simultaneously with the retinal evoked response, to consider the possibility for changes in the cortical response. The data do not support the second hypothesis (H2B), because paired t-tests revealed no change in PVER amplitude from one visit to the other (p > 0.05). Rather, steady-state PVER amplitude was reduced at both test visits in the
143 post-migraine period (Figure 3.4C; main effect of group: p = 0.002, interaction between group and visit: p = 0.45). Thus, despite the presence of visual field defects that were worse after migraine, deficits in the cortical evoked response remained stable after migraine.
The implications of, and possible mechanisms underlying these results are discussed in the following sections.
3.4.1 Generalised visual field changes in the days after migraine
Previous work implies that generalised deficits in visual field sensitivity are worse the day after migraine, based on group comparisons of the Average Defect (McKendrick and Badcock 2004b). Likewise, this study found that there was a significant interaction between group and visit for the Average and Mean Defect indices across the three perimetric tasks. Paired t-tests showed a significant improvement in SAP Average Defect in the control group only (Figure 3.6A), suggesting that people with migraine performed worse than controls at the second visit, which was conducted within six days of a migraine. Alternatively, as SAP was the first perimetric test conducted for all participants, these results can also be interpreted as a lesser learning effect in the migraine group. The trend for worse generalised sensitivity at the visit closer to a migraine was not significant when considering the interaction between group and visit for TMP (Figure 3.6B) and SWAP (Figure 3.6C). However, some migraine individuals showed a reduction in Average/Mean Defect that was worse than the 2nd worst performing control participant (Figure 3.7).
A reduction in overall visual field sensitivity, or lesser learning effect immediately post-migraine, is possibly explained by fatigue or poor concentration as a result of anti-migraine medications or the symptoms of migraine itself. However, post- migraine effects were minimised by scheduling test visits at least one day after the offset, and not onset, of all migraine symptoms. Upon questioning, all participants felt normal at the second test visit immediately after a migraine. An alternative reason
144 for reduced sensitivity is aversion to the test stimuli (Marcus and Soso 1989). This might also explain the reduction in steady-state PVER amplitude, as discussed in Chapter 2. Aversion was not formally measured; however, as per the previous experiment (Chapter 2), none of the participants reported discomfort or voluntarily withdrew from the study during testing. Furthermore, participants with a strong aversion to visual stimuli may have excluded themselves from volunteering for a study that explicitly involved extensive visual testing for up to three hours.
An underestimation of visual field sensitivity can occur if a person does not respond to a visual field stimulus that was previously ‘seen’, i.e. false-negative. However, it is unlikely that worse generalised sensitivity in the migraine group at the second visit was due to missed button presses. The perimeters do not provide information regarding response times, but studies show that most people respond within 300 to 450 ms on standard visual field testing (Wall et al. 1996). People with migraine demonstrate an approximate 30 ms increase in reaction times in the 72 hours after an attack (Mazzucchi et al. 1988; Siniatchkin et al. 2006). This amounts to a 10% delay in the normal response time for visual field tests. Such a delay in button pressing is unlikely to affect visual field performance, as the perimeter allows a response window of at least 1000 ms before the next stimulus presentation. Moreover, average false negative rates were not different between groups (see Appendix B, Section 8.4.1), which would have been expected in the presence of severely delayed responses. A high false-negative rate does not necessarily imply unreliable performance, but can result from a flatter psychometric function in areas of depressed visual field (Artes et al. 2002), which is consistent with the migraine group performance.
3.4.2 Localised visual field changes in the days after migraine
In contrast to the study by McKendrick and Badcock (2004b), there was no change in Pattern Defect and Loss Variance in the days immediately after migraine (Table 3.6; paired Wilcoxon signed rank tests, p > 0.05). The discrepancy between the two
145 studies might be due to the timing of testing. Those participants were tested one day after migraine offset, with test sessions lasting no more than one hour (McKendrick and Badcock 2004b). The participants in this study, however, were asked to return as soon as practicable but within one week of a migraine, as it was anticipated that it would be more difficult to arrange a second visit of three hours duration at short notice. As a result, the average time post-migraine at the second visit was not 24 hours, but three days (Table 3.1). The more demanding nature of the long test session likely biased the timing of the second visit to a day further away from a headache, and prevented migraine participants from completing the second session the day after an attack, where performance was likely to be worse. Thus, taken together, the literature suggests that localised changes in the visual field appear to be most pronounced the day after a migraine.
A significant number of locations with a more pronounced reduction in sensitivity were identified within six days of a migraine. A well-recognised limitation is that visual field data are variable in areas of visual field loss (Artes et al. 2002). Quantification of test-retest performance is often used to address this issue (e.g. McKendrick and Badcock 2004b). However, it was not practicable to measure visual fields more than once on the same day to estimate the reliability of the changes in visual field sensitivity in the migraine participants, due to the long test protocol. Moreover, collecting test-retest data for people with migraine on different days is confounded by potential differences in performance, as visual field sensitivity is known to vary with time relative to a migraine (Drummond and Anderson 1992; Sullivan-Mee and Bowman 1997; McKendrick et al. 2000; McKendrick and Badcock 2004b; Goodwin 2011). The point-wise analysis was therefore designed to measure sensitivity in the migraine group on two occasions and evaluate this against the variability found in control participants, who were also tested on two separate days. It was found that the decrease in sensitivity was not due to increased variability, as the upper limits of test-retest performance in both groups indicate a similar number and degree of relatively improving locations (Figure 3.11), consistent with a previous report (McKendrick and Badcock 2004b).
146 That sensitivity at discrete locations in the visual field was worse in the days after a migraine suggests that these changes are not a result of fatigue or aversion, which would be expected to produce an overall reduction in sensitivity. Rather, these changes might be sequelae of the migraine event. Although the mechanism for localised deficits in visual field sensitivity after migraine cannot be ascertained from this study, it has been suggested that this might result from localised vascular events (Flammer et al. 2001). Such events could plausibly occur at the brain or peripherally. Temporary reductions in regional cerebral blood flow in the visual cortex of people with migraine have been found during attacks using brain neuroimaging techniques (Woods et al. 1994; Cao et al. 1999; Hadjikhani et al. 2001; Denuelle et al. 2008), which could plausibly cause focal visual deficits at the visual cortex. In any case, fMRI and PET measures of regional cerebral blood flow do not have the same spatial resolution as the visual field measures employed in this study. Moreover, whether the vascular compromise reaches a level consistent with ischaemia remains unresolved (see Chapter 1).
Alternatively, studies suggest that people with migraine have poor peripheral vascular regulation (Zahavi et al. 1984; Gasser and Meienberg 1991; Hegyalijai et al. 1997; Flammer et al. 2001). Of particular interest to this thesis is the association between migraine and transient retinal vasospasms, which have been visualised in people with migraine (Killer et al. 2003; Abdul-Rahman et al. 2011; Flammer et al. 2013), which could plausibly cause altered perfusion and increase the risk of focal ischaemic damage to the optic nerve head (Gasser and Meienberg 1991; Flammer et al. 2001) and retina (Gutteridge et al. 2007). This mechanism is presumed to underlie visual deficits in patients with normal tension glaucoma (NTG), who demonstrate reduced blood flow at the optic nerve head (Plange et al. 2003). Hence, altered peripheral vascular perfusion in migraine may increase the risk of NTG in these individuals and explain the development of monocular visual field loss in the days immediately after a migraine (e.g. Figure 3.12B). Indeed, people with focal ischaemic-type optic nerve head damage from NTG have a higher prevalence of migraine (Broadway and Drance 1998).
147 Using point-wise analysis, only three people with migraine (19%) showed normal results for every visual field task at every visit, compared with 20 of 26 control participants (77%). The percentage of visual field deficits in migraine sufferers in this study was higher than that reported in other studies, as a participant could be flagged as abnormal on least one of the three visual field tasks implemented, i.e. SAP, TMP, and SWAP. For example, Lewis et al. (1989) found 35% (21 of 60) and De Natale et al. (1993) found 42% (17 of 38) of their migraine participants showed visual field deficits using SAP. In contrast, studies using perimetric stimuli that preferentially bias the M or K pathway – TMP and SWAP, respectively – frequently report larger proportions of visual field defects in their migraine cohorts, e.g. 67% (10 of 15) on TMP (McKendrick et al. 2000) or 53% (8 of 15) on SWAP (Yenice et al. 2006). Hence, collection of data using a number of perimetric tasks was likely to increase the likelihood of finding a visual field defect in this study, with different people with migraine showing visual field defects on different tasks. That SAP, TMP, and SWAP defects were found is consistent with other evidence of non-neural pathway specific pre-cortical deficits in migraine (Coleston et al. 1994; McKendrick and Badcock 2003; McKendrick and Sampson 2009).
3.4.3 Electrophysiological changes in the days after migraine
This study is the first to evaluate changes in the retinal electrophysiological response with time post-migraine, using the PERG. Previous visual field studies find a high proportion of individuals with monocular deficits that appear retinal in origin (Lewis et al. 1989; De Natale et al. 1993; McKendrick et al. 2000; McKendrick et al. 2002; Çomoglu et al. 2003; McKendrick and Badcock 2004a). Although some visual field defects can be localised and even appear arcuate in pattern (e.g. Figure 3.12B), there is also evidence for generalised sensitivity losses in one eye that are worse the day after migraine (McKendrick and Badcock 2004b). Hence, as per the previous experiment, the PERG was measured in response to a 30° checkerboard stimulus, which encompassed the spatial extent of all the perimetric tasks included in this study. All PERG parameters were consistently within normal limits (Table 3.2),
148 implying no diffuse retinal dysfunction in migraine and no changes in generalised retinal function after migraine. This is in contrast to the PERG in glaucoma, which is often reduced in amplitude (Price et al. 1988; Bach and Speidel-Fiaux 1989; Parisi et al. 2006; Porciatti and Ventura 2009b) and delayed in timing (Price et al. 1988; Parisi et al. 2006; Porciatti and Ventura 2009b) with little correlation to visual field defects (Bach et al. 1992; Hood et al. 2005), suggesting a diffuse retinal involvement in glaucoma that is not present in migraine. This is discussed later in the final chapter of this thesis.
Unlike the visual field tests, the PERG technique employed was not designed to find small, localised losses or investigate the spatial extent of visual dysfunction. A potential tool in electrophysiology for spatial localisation is the use of a multifocal technique (Hood et al. 2012). For multifocal PERG, the stimulus field is divided into smaller elements scaled for retinal eccentricity, e.g. ranging from 2° to 6° in size (Klistorner et al. 2000; Harrison et al. 2006). Thus, this method does not have the same spatial resolution as the visual field tests employed in this study, which identified sensitivity losses in people with migraine at discrete locations using test stimuli of 0.5° (SAP, TMP) and 1.7° (SWAP), although visual field defects have also been found using larger 10° motion targets (McKendrick and Badcock 2004c). Nevertheless, placement of visual stimuli at different locations in the visual field might uncover PERG deficits that are spatially localised. It is worth noting, however, that reductions in the multifocal PERG response do not necessarily correspond to locations of visual field loss in people with glaucoma (Klistorner et al. 2000; Harrison et al. 2006).
To date, studies have only measured the transient PVER at multiple time-points post-migraine, whereas changes in the steady-state PVER after migraine have not been investigated. Judit et al. (2000) and Sand and Vingen (2000) divided participants into groups based on the number of days after migraine. Deficient PVER habituation (Coppola et al. 2009) was not different when measured at one day, two days, and more than three days after a migraine (Judit et al. 2000). These findings are not directly comparable to the current findings, as PVER habituation was not specifically
149 measured here (see discussion in Chapter 2). In the study by Sand and Vingen (2000), the PVER amplitude of people with migraine who happened to be tested within 24 hours of an attack were not different from the amplitude of those people tested at least three days after migraine. Thus, the results of this experiment are consistent with previous reports (Judit et al. 2000; Sand et al. 2008; Sand et al. 2009), finding a consistent defect in the cortical response that appears unrelated to the timing of the test after a migraine.
In contrast, changes in the transient PVER after migraine have been reported by Shibata and colleagues, where testing was limited between one and 30 days after migraine. This study found that PVER amplitude was most abnormal in the subgroup of people tested within 10 days after migraine, compared to 11 to 20 and 21 to 30 days after migraine (Shibata et al. 1998). Similar subgroup analysis could not be performed here due to differences in experimental design. The aim of this study was to test people with migraine at two visits to allow within-individual comparisons, which was not done by Shibata and colleagues. Furthermore, we ensured that the second visit (one to six days post-migraine) was scheduled at a time that was quite different from the baseline visit (between seven and 150 days after a migraine, Figure 3.1), relative to the last migraine. Shibata et al. (1998) assumed that the people who were tested 21 to 30 days after an attack were likely to have been tested just before another attack; thus, the authors suggested that the change in PVER amplitude over time reflected changes in cortical excitability in the lead up to another migraine. However, actual data regarding time pre-migraine were lacking. Hence, it is unknown whether the gradual ‘normalisation’ of cortical responses over time is a feature of the migraine cycle or an effect of the migraine event itself. Future experiments that involve daily measurement of visual function between and during migraine events can address these hypotheses. The issue is partly resolved by the PVER studies discussed in Chapter 1 that show prominent changes in cortical activity before and during an attack, yet insignificant changes after an attack (Judit et al. 2000; Sand et al. 2008; Sand et al. 2009). This suggests that changes in neural activity identified using electrophysiology are related to cortical susceptibility, which coincide with prevailing theories of normalisation of cortical excitability contributing to the triggering of a
150 migraine attack (Chen et al. 2009; Siniatchkin et al. 2009). In particular, cortical excitability is believed to predispose people to migraine (Vecchia and Pietrobon 2012) and has been inferred from neurophysiological tests and psychophysical tests of visual perception (for reviews, see Aurora and Wilkinson 2007 and Coppola et al. 2007c). Given that in the first two experiments of this thesis (Chapters 2 and 3) the steady-state PVER amplitude was abnormal in people with migraine, the next chapter explores the relationship between the steady-state PVER and psychophysical measures of cortical excitability.
The steady-state but not transient PVER was abnormal in the migraine group. Steady-state electrophysiological responses have consistently demonstrated differences in people with migraine (Diener et al. 1989; Marrelli et al. 2001; Shibata et al. 2008). In some cases, as observed here, a clear separation between migraine and control groups was only measurable in the steady-state response (Marrelli et al. 2001). It is possible that the stable steady-state PVER deficits reflect an epiphenomenon of migraine, rather than related to susceptibility to migraine. Flicker is known to induce higher metabolic demands and increase blood flow in the brain (Pastor et al. 2003). The stimulation rate used for steady-state recording in this study was approximately 8 Hz, which corresponds to the temporal frequency that normally produces a maximal change in the cerebrovascular response to a flickering checkerboard pattern, as measured by fMRI-BOLD (Singh et al. 2003). Alternatively, the reduction in steady- state PVER amplitude might reflect abnormal visual motion processing, given that the major sources of the steady-state PVER are cortical areas V1 and V5/MT (Di Russo et al. 2007). As discussed in Chapter 1, there is converging evidence for altered function and structure in MO and MA groups in the visual motion processing pathways from studies using TMS (Battelli et al. 2002), neuroimaging (Granziera et al. 2006; Antal et al. 2011), and behavioural measures of global dot motion (McKendrick and Badcock 2004c; Antal et al. 2005; Ditchfield et al. 2006; McKendrick et al. 2006a; McKendrick et al. 2006b; Shepherd et al. 2012), which could have a secondary effect on the response measured at the visual cortex by PVER.
151 3.4.4 Anatomical locus of visual anomalies in migraine
Both the electrophysiological and visual field tests employed in this study enable inferences about whether visual deficits in migraine are likely to arise from cortical or retinal sources. Visual field defects were classified on the basis that monocular patterns of field loss arise from pre-chiasmal or retinal dysfunction, whereas the homonymous nature of migraine visual aura and bilateral field loss are supportive of a cortical origin. In this study, there were small, localised regions of reduced sensitivity that improved with time after a migraine, whereas bilateral visual field defects resembling cortical dysfunction tended to be stable and involved larger and more generalised regions. The latter observation is consistent with the abnormal PVER amplitudes at both test visits without a corresponding deficit in the PERG, which collectively imply a generalised cortical and not retinal dysfunction in migraine.
3.5 Conclusions
This study is the first to show that retinal and cortical electrophysiological responses do not correlate with deficits in visual field performance in the same individuals with migraine. On the one hand, electrophysiological tests find a generalised cortical anomaly in the migraine group. On the other hand, visual field tests find patterns of defect consistent with retinal and cortical dysfunction. The lack of correlation may be related to the spatial extent of the test targets or the timing of the changes observed. The electrophysiological techniques identify diffuse dysfunction, whereas visual field tests are capable of detecting small, localised losses in sensitivity that are worse in the days immediately after a migraine.
Visual field defects are hypothesised to represent adverse sequelae of migraine, as they are worst in the days following an attack. In contrast, deficits in the PVER remained stable after a migraine. Without having measured visual function at multiple times (before, during, and after a migraine), it is presumed that these cortical
152 deficits are related to predisposition to migraines. A possible mechanism underpinning migraine susceptibility (the imbalance between cortical inhibition and excitation), which is manifest in both neurophysiological and perceptual responses, is investigated in the final experiment of this thesis (Chapter 4).
153
154 4 | Chapter Four Imbalance between excitation and inhibition in visual processing in migraine
4.1 Introduction
In the previous chapters, the PVER was used to infer neural activity in the visual cortex. The PVER abnormality found in Experiment 1 (Chapter 2) was not related to a deficit in the retinal response (PERG) and hence was interpreted as originating at the level of the cortex. Neither was the PVER abnormality significantly changed after a migraine event (Chapter 3), which implied that the cortical anomaly was not a direct consequence of migraine. Previous literature suggests that PVER abnormalities gradually build up in intensity and then normalise at the onset of an attack (Judit et al. 2000; Sand et al. 2008; Sand et al. 2009). Thus, the PVER abnormality found in the previous chapters could reflect susceptibility to migraine. One of the competing theories regarding migraine susceptibility is an imbalance between inhibition and excitation, which has been proposed as the putative basis for cortical hyperexcitability in migraine (Vecchia and Pietrobon 2012). As discussed in Chapter 1, cortical hyperexcitability could be attributed to excessive excitation or weakened inhibition. However, whether intracortical excitation or inhibition, or both, is affected in migraine is still unclear.
One possible consequence of hyperexcitability is increased spontaneous neural firing. The level of spontaneous neural activity has more recently been inferred in two perceptual studies (Wagner et al. 2010; Webster et al. 2012), by estimating the level of internal noise and effect of external noise on psychophysical measures of contrast sensitivity. Both of those studies found increased internal noise, which was scaled by
155 the magnitude of the stimulus (i.e. multiplicative noise), implying a contrast- dependent elevation in neural excitation (Wagner et al. 2010; Webster et al. 2012). However, these studies were unable to consider whether inhibition might also be affected.
Other visual perceptual studies have specifically used tasks that are considered to rely on cortical inhibitory function, with varied findings. Reduced masking (Palmer et al. 2000) and impaired orientation discrimination (McKendrick et al. 2001; Tibber et al. 2006) imply a reduction in inhibition. However, studies that find an increase in the strength of after-effects in people with migraine argue that such perceptual anomalies are not readily explained by reduced inhibition (Shepherd 2001; Shepherd et al. 2002; Shepherd 2006). In contrast, Battista et al. (2010) also measured the motion after- effect and found the opposite result, i.e. weaker after-effects that imply an increase in inhibition. Still others find no significant difference in the strength of inhibition (McColl and Wilkinson 2000; Wilkinson and Crotogino 2000).
In electrophysiological studies in cat, the effect of altered inhibition on the visual evoked response has been demonstrated by injection of the GABA antagonist, bicuculline, which was found to increase the amplitude of the VER (Zemon et al. 1980). As the VER is a summation of excitatory and inhibitory post-synaptic potentials (Eccles 1951), an increase in the cortical response could reflect increased excitation or decreased inhibition. That people with migraine show increased PVER amplitude (for reviews, see Ambrosini et al. 2003 and Magis et al. 2007a) and progressively larger PVER amplitudes in response to continuous stimulation, also known as deficient habituation (reviewed in Coppola et al. 2009), is suggestive of a generalised inhibitory dysfunction (Palermo et al. 2011). In this model, reduced inhibition is inconsistent with the smaller PVER amplitude observed thus far in this thesis (Chapters 2 and 3) and in previous studies of the PVER (Polich et al. 1986; Khalil et al. 2000; Boylu et al. 2010). Hence, PVER studies have provided inconclusive evidence regarding whether excitation or inhibition is affected, because of the difficulties in disentangling effects of excitation and inhibition in the absence of invasive procedures and pharmacologic intervention.
156 An alternative approach that has been used to differentiate between changes in excitation and inhibition is to measure perceptual analogues of centre-surround effects. This method has been utilised to study various clinical disorders believed to be due to altered brain excitability (Dakin et al. 2005; Tadin et al. 2006; Yoon et al. 2010), including migraine (Battista et al. 2010; Battista et al. 2011). Whether perceptual suppression or facilitation occurs depends on the balance between excitatory and inhibitory connections within and to V1 (Angelucci and Bressloff 2006; Schwabe et al. 2010). As described in Chapter 1, perceptual surround suppression can be elicited using a number of different stimuli. In the most recent study by Battista et al. (2011), the contrast-contrast task, or Chubb illusion (Chubb et al. 1989) was implemented. When a central patch of grating of 40% contrast was presented with a surrounding annular stimulus of similar pattern but higher contrast (95%), both migraine and control groups demonstrated a reduction in the perceived contrast of the central stimulus. People with migraine showed relatively more perceptual suppression (Battista et al. 2011). This result is consistent with the reduction in PVER amplitude found in Chapters 2 and 3, but appears to conflict with the notion of hyperexcitability in migraine and with psychophysical findings that can be interpreted as arising from impaired cortical inhibitory mechanisms (Palmer et al. 2000; McKendrick et al. 2001; Tibber et al. 2006; Wagner et al. 2012).
Based on our current understanding of the neurophysiology involved in centre- surround antagonism (Angelucci and Bressloff 2006; Schwabe et al. 2010) (see Figure 1.5 of Chapter 1), there are at least two ways that increased perceptual suppression in people with migraine could come about, as hypothesised by Battista and colleagues (Battista et al. 2010; Battista et al. 2011). Firstly, given that suppressive effects are believed to arise from lateral inhibition within V1, increased suppression could be due to a primary increase in inhibitory drive but with no alteration in the overall excitatory drive. As outlined earlier, it is still unclear whether inhibition is abnormal in migraine. Alternatively, a more likely explanation is that the lateral inhibitory networks are also innervated by feedback connections from extrastriate visual areas, which are known to be excitatory (Angelucci and Bressloff 2006; Schwabe et al. 2010). Thus, an overall increase in excitation could have a secondary effect of
157 increasing the amount of lateral inhibition, which would dominate the local excitatory effect. This explanation could also account for the reduced PVER amplitudes in the previous experiments (Chapters 2 and 3). Thus, in this final experiment, perceptual suppression and the PVER were measured in the same individuals.
An important feature of both of these methods is that the net response reflects the balance between excitation and inhibition, which in turn shows a strong dependence on contrast. The perceived contrast of a central stimulus is normally reduced in the presence of a high contrast surround (Chubb et al. 1989), as would be expected from surround inhibition. However, as the contrast of the centre decreases, the suppressive effect increases with a fixed contrast of the surround (Ejima and Takahashi 1985; Xing and Heeger 2001; Yu et al. 2001; Karas 2012). Similarly, greater suppression is observed when the contrast of the surround increases under conditions of fixed centre contrast (Ejima and Takahashi 1985; Xing and Heeger 2001; Yu et al. 2001; Karas 2012). Furthermore, when the contrast of the centre exceeds that of the surround, there can be perceptual facilitation, rather than suppression (Ejima and Takahashi 1985; Xing and Heeger 2001; Yu et al. 2001; Karas 2012). Thus, the contrast ratio between centre and surround is a critical factor in determining the amount of suppression or facilitation. If there were an overall increase in inhibition in migraine affecting the lateral inhibitory connections, perceptual suppression would occur independent of contrast. Alternatively, if inhibition were a secondary effect of increased excitation, people with migraine would be expected to show the most perceptual suppression under conditions that normally elicit perceptual suppression, which leads to the first hypothesis:
H3A: That perceptual surround suppression in people with migraine is increased for centre-surround stimuli of lower contrast centre and higher contrast surround.
Contrast is also an important variable determining the amplitude of the PVER. The cortical contrast response function, or contrast gain, is known to be non-linear.
158 Single cell studies of visual cortical neurones (Albrecht and Hamilton 1982; Sclar et al. 1990) and PVER measurements in humans (e.g. Spekreijse et al. 1973; Ross and Speed 1991; Porciatti et al. 2000; Heinrich and Bach 2001; Tsai et al. 2011) typically show a saturating response, which has been explained using the ‘divisive normalisation’ model by Heeger (1992). This model incorporates the interaction between excitatory and inhibitory inputs, whereby a neurone’s output is driven by excitation of its receptive field but also excitation of an inhibitory pool of neurones in the surround (the ‘normalisation’ pool). At the highest contrasts, excitation is increased, which drives the primary neuronal response and also activates inhibitory interneurones via feedback connections, reducing the net excitatory output (Heeger 1992; Carandini et al. 1997). If there were an overall increase in inhibition in migraine affecting the primary neuronal response, then the contrast gain would still show saturation, but smaller responses relative to the normal response at each contrast level. Alternatively, if inhibition were a secondary effect of increased excitation, then the cortical evoked response would be expected to be higher than normal at low contrasts, but would be relatively dampened by excessive inhibition from the ‘normalisation pool’ at high contrasts. This response decline at high contrasts, or ‘supersaturation’, has been demonstrated in 25 to 35% of cortical neurones (Li and Creutzfeldt 1984; Ledgeway et al. 2005) and in some human PVER contrast response functions (Tyler and Apkarian 1985; Morrone and Burr 1986; Burr and Morrone 1987; Tsai et al. 2012). Thus, the second hypothesis of this study was:
H3B: That the cortical electrophysiological response in people with migraine is increased for stimuli of lower contrast and decreased for stimuli of higher contrast.
This study also considered whether the contrast-dependent changes in PVER were consistent with the notion of hyperexcitability in migraine by comparing the shape of the contrast response function. Firstly, the saturating non-linearity defined by Equation 4.1 describes most cortical neural responses ( R) as a function of contrast ( c):
€ € 159 c n R(c) = Rmax ⋅ n n + R0 (Equation 4.1) c + c50
where R0 is the inferred internal noise returned from the response at 0% contrast, € Rmax is the maximum response, c50 is the semi-saturation point, which represents the contrast at which the response is halfway between baseline and maximum, and n is € the excitatory exponent, which determines the positive slope of the function. € € Examples of saturating contrast response functions are shown by the solid curves in € Figures 4.1 and 4.2.
To account for supersaturation, Peirce (2007) included an additional parameter s to Equation 4.1:
€ c n R(c) = Rmax ⋅ sn sn + R0 (Equation 4.2) c + c50
The characteristics of Equation 4.2 are similar to Equation 4.1 in that the exponents, € n and s, control the shape of the function, where n reflects excitatory components
and s reflects inhibitory components. Rmax is the ‘projected’ maximum response as predicted from the rising limb and not the maximum point of the function, and c € € € 50 is the contrast at which the response is halfway between baseline and the projected € € maximum. The inverse of c50 is the sensitivity of the excitatory process. A value of € s > 1 represents supersaturation, provided that the other exponent n is fixed. Note that when s has a value of 1, Equation 4.2 reduces to Equation 4.1 and implies a € saturating response. In this study, it was assumed that the normal PVER contrast € € gain conforms to a saturating non-linearity, as described by Equation 4.1. If € increased perceptual suppression arises from a primary dysfunction of increased inhibition, where excitation remains unaltered, this would predict a uniform decrease in the PVER across all contrast levels. Such changes are simulated in Figure 4.1,
based on changes to the Rmax and c50 parameters of Equation 4.1. These predictions, however, do not support the hyperexcitability state reported in migraine.
€ € 160
Figure 4.1 – Simulated changes in normalised PVER contrast gain that result in a reduction in PVER amplitude across all stimulus contrasts, presuming a divisive normalisation model of Heeger (1992). The solid curves indicate the normal response.
The dashed curves indicate the changes to each free parameter ( Rmax and c50) of
Equation 4.1, whilst keeping the other constant. (A) Reduced Rmax (B) Increased c50.
The vertical dotted lines indicate the semi-saturation constant, or c50. All other € € parameters were fixed ( n = 1.4, R0 = 0.14, see Section 4.2.3.3). € €
€ Alternatively, if there were excessive excitation of lateral inhibitory connections, one € € would expect to find a relatively greater inhibitory effect and therefore a decline in the response at high contrasts, i.e. supersaturation. Figure 4.2 depicts the anticipated changes depending on whether the remaining free parameters, Rmax and c50, are also altered, whilst s is greater than 1. That is, in all possible scenarios, the dashed lines show an initial increase in the response and then a relative decline with increasing € € contrast. €
161
Figure 4.2 – Simulated changes in normalised PVER contrast gain that result in a relative response decline at the highest stimulus contrasts (supersaturation). The solid curves indicate the normal response, where s = 1.0. The dashed curves ( s > 1) indicate the
changes to each free parameter ( Rmax and c50) of Equation 4.2, whilst keeping the other
constant. (A) Normal Rmax and c50, showing a reduction in excitation at low contrast (B) € € Decreased c50, showing normal excitation and inhibition at low contrast (C) Increased € € Rmax , showing increased excitation at low contrast. The vertical dotted lines indicate the € € semi-saturation constant, or c50. All other parameters were fixed ( n = 1.4, R0 = 0.14, see€ Section 4.2.3.3). € € Figure 4.2 shows €that the balance between excitation and inhibition€ at low contrasts can vary depending on the manipulation. Figure 4.2A shows the scenario where there is a reduction in the response at all contrasts. This is similar to Figure 4.1 that depicts an overall reduction in the level of excitation, except that there is a further decline in the response at high contrasts. Figure 4.2B is an example of a supersaturating
function that yields normal excitation at low contrasts if c50 is decreased. In the same way as Figure 4.1, Figures 4.2A and 4.2B fail to describe the hyperexcitable state reported in migraine, as the response at low contrasts is normal or reduced. € Alternatively, if Rmax is increased, the response shows increased gain before supersaturation takes place (Figure 4.2C). This suggests a hyperexcitability that is evident at low contrasts, in addition to the excessive inhibition at high contrasts. €
162 4.2 Methods
4.2.1 Participants
A power analysis was performed using data from studies that have identified differences in people with migraine with the specific techniques employed here. Psychophysical data were taken from a study that previously measured perceptual centre-surround suppression using a drifting contrast-contrast task (Battista et al. 2011). In addition, the steady-state PVER results from Experiment 2 (Chapter 3) revealed a significant decrease in amplitude in both MO and MA participants at both test visits. Hence, amplitudes were averaged across the two visits to obtain population estimates for each group (mean ± standard deviation: controls 3.8 ± 1.9 µV, migraine 1.9 ± 1.2 µV). The analyses indicated that 10 participants in each group provided a power of at least 0.80 (α = 0.05) for detecting a 20% increase in perceptual suppression or 50% decrease in steady-state PVER amplitude in people with migraine.
Twenty non-headache controls and 24 migraine participants (12 MO, 12 MA) were recruited after meeting the eligibility criteria detailed in Chapter 2. There was no difference between MO and MA participants in their migraine characteristics (Table 4.1). This varies from the cohort recruited in the previous experiments (Chapters 2 and 3) where MIDAS questionnaire scores were higher in the MO group. In addition, subgroup analysis confirmed similar perceptual centre-surround responses and steady-state PVER in both forms of migraine (see Appendix C, Section 9.4.1), consistent with previous studies (Shibata et al. 2008; Battista et al. 2010; Battista et al. 2011). Hence, the MO and MA participants were pooled. The control (aged 21 – 47, mean ± standard deviation: 29 ± 7 years) and migraine groups (aged 20 – 46, 31 ± 8 years) were of similar age (t42 = 1.01, p = 0.32). The test visit was scheduled at least three days after a migraine, and at least three days before the next migraine.
163 Table 4.1 – Self-reported migraine characteristics (median, minimum – maximum).
Migraine characteristic MO MA Statistic p-valuea
Days since last migraine 13 (5 – 100) 16 (4 – 104) U = 71.0 0.98
Age at first migraine (years) 12 (9 – 25) 13 (10 – 32) U = 54.0 0.31
Migraine history (years) 16 (8 – 30) 11 (1 – 34) U = 63.5 0.64
Migraines in past year 12 (3 – 40) 6 (3 – 50) U = 66.5 0.77
Weeks between migraines 4 (2 – 24) 8 (1 – 12) U = 68.0 0.84
Estimated number of lifetime attacks 129 (60 – 360) 50 (14 – 1350) U = 50.0 0.21
MIDAS questionnaire score (days) 8 (0 – 32) 5 (0 – 41) U = 58.0 0.43 a p-values represent the significance of group comparisons (Mann Whitney rank sum or t-tests)
4.2.2 Perceptual centre-surround task
4.2.2.1 Equipment and room setup
The centre-surround stimuli were created using custom written software in Matlab 7.0 (Mathsworks, Natick, MA, USA) and presented using the ViSaGe system (Cambridge Research Systems, Kent, UK) on a Sony G520 21-inch CRT monitor (Sony Corporation, Tokyo, Japan) with a resolution of 600 × 800 pixels and a frame rate of 100 Hz. The monitor was turned on at least 30 minutes prior to testing and calibration (Metha et al. 1993). Monthly automated gamma corrections were performed using the Optical photometer (Cambridge Research Systems, Kent, UK), the details of which appear in Appendix C, Section 9.1. Participants were seated comfortably with the head stabilised by a forehead and chin-rest, with the appropriate refractive correction for the 90 cm working distance.
4.2.2.2 Stimulus characteristics
The centre (0.67° radius) and surround (4° radius) patterns were in-phase vertical sinusoidal gratings with a spatial frequency of 2 cycles/degree and a drift rate of 2
164 degrees/second, the direction of which (leftwards or rightwards) was randomised per trial. The centre was either presented alone (‘no surround’ condition) or with the presence of the annular surround (‘surround’ condition). Example stimuli are shown in Figure 4.3. Stimuli were identical to that used previously by Battista et al. (2011), with two exceptions. Firstly, the contrast of the centre and surround varied (20, 40, 60, 80%) in order to investigate the contrast-dependence of perceptual responses (H3A). The second difference was the presence of a 0.1° gap between the centre and surround (Figure 4.3). The gap was inserted in order to enable identification of the centre and surround when the two were of equal contrast, and was present for all stimulus presentations to keep any effects of the gap consistent.
Figure 4.3 – Example stimuli and schematic of the two-interval forced choice (2IFC) task used in the perceptual experiment. The first interval (500 ms) contained the centre alone (0.67° radius, ‘no surround’ condition) of fixed contrast (20, 40, 60, or 80%), followed by an interstimulus interval (500 ms) of a blank screen. The second interval (500 ms) contained a central target of variable contrast surrounded by an annulus (‘surround’ condition) of fixed contrast (20, 40, 60, or 80%). The centre and surround were gratings (2 cycles/degree) drifting at 2 degrees/second. A 0.1° gap was inserted to enable differentiation between the centre and surround. The task was to identify which interval contained the central target of higher contrast.
165 The order of the 16 centre-surround combinations was randomised between observers and counterbalanced between groups. Previous work demonstrates that central flicker contrast detection thresholds are not different between migraine and control groups (McKendrick et al. 1998; Karanovic et al. 2011); hence, all observers were capable of seeing the target at its minimum contrast of 20%, as confirmed during testing by verbal feedback.
4.2.2.3 Thresholding algorithm
The ‘no surround’ condition was always presented first, which was of fixed contrast, followed by the ‘surround’ condition (Figure 4.3). The aim of the task was to vary the contrast of the ‘surround’ condition (second interval) to match that of the ‘no surround’ condition (first interval). A two-interval forced choice (2IFC) paradigm was used to determine the matching contrast, where participants indicated the interval that contained the central stimulus of higher contrast, by means of a button press (button box model CB6, Cambridge Research Systems, Kent, UK). No feedback was provided. Participants completed practice trials in order to familiarise themselves with the task. The tests were self-paced and lasted up to two hours, with rest breaks as required.
The thresholding procedure was a single 1-up, 1-down staircase using 10% step sizes, which commenced at a contrast level that was supra-threshold. Two repeats were averaged for each stimulus condition. One run was started above whereas a second run was started below the veridical contrast (Figure 4.4). In the example shown in Figure 4.4 by filled symbols, the central stimulus was fixed at the veridical contrast of 40% (first interval). The second interval was a stimulus of higher contrast (60%); hence, the matching contrast was decreased by a 10% step in contrast (54%). This procedure continued until the opposite interval was chosen (i.e. one reversal), at which point the matching contrast was increased by the same step size (10%). The staircase terminated after ten reversals, with the average of the last six reversals being
166 used as an estimate of perceived contrast for each run. The example in Figure 4.4 shown by the unfilled symbols demonstrates a staircase beginning at 20% contrast.
Figure 4.4 – Example 1-up, 1-down staircase results for an 80/40% centre/surround stimulus, using step sizes of 10%. The veridical contrast of the centre was 40% (dotted horizontal line). The matching contrast was varied, beginning at 60% (“above”, filled symbols) or 20% (“below”, unfilled symbols). When the first interval was chosen as containing the higher contrast stimulus, the matching contrast was increased by 10%, or vice versa, until the opposite interval was chosen (first reversal). The staircase terminated after 10 reversals, with the average of the last six reversals being the estimated perceived contrast of each run. The final estimate of perceived contrast was the average from the two runs (i.e. 2 x 6 reversals).
A ‘suppression ratio’ was calculated (SR = perceived contrast / physical contrast) to quantify the effect of the surround on the perceived contrast (Battista et al. 2010; Battista et al. 2011), where a ratio of less than one indicated suppression and a ratio greater than one indicated facilitation.
167 4.2.3 Pattern electrophysiology
4.2.3.1 Monocular testing
Although psychophysical tests are often conducted binocularly, monocular stimulation for PVER recordings is the standard (Odom et al. 2010) and was therefore the approach adopted throughout this thesis. For this experiment, a single eye was chosen at random to enable comparisons with the result of previous chapters. A translucent patch covered the other eye. The same eye was used for both perceptual and electrophysiological tests.
4.2.3.2 Recording methods
The steady-state PVER was recorded to assess cortical function as per the methods described in Chapter 2 with the exception of two major differences. Firstly, unlike the previous chapters, only the PVER was measured in this experiment. Hence, of the five possible recording channels on the Espion system (Figure 4.5A), only one was required to record the PVER in response to the main gold cup electrode placed at OZ (Figure 4.5B). OZ is typically located at 2.5 cm above the inion and is presumed to overlie the location of the visual cortex within the calcarine fissure (Jasper 1958). This is the standard electrode position for PVER recordings (Odom et al. 2010). However, based on fMRI scans, the calcarine fissure ranges from 2 cm below to 3.5 cm above the inion (Hood and Zhang 2000; Ishikawa et al. 2011). These anatomical differences are believed to increase the inter-individual variability in PVER responses (Brindley 1972; Steinmetz et al. 1989). Thus, in this study, two extra gold cup electrodes were placed on the scalp along the midline at 5% nasion-inion distance above and below OZ, as illustrated in Figure 4.5C. The final PVER estimate was the average response across these three electrode positions.
168
Figure 4.5 – Schematic of the electrode placement for PVER recording. (A) Three of the five recording channels were utilised, each consisting of an active (positive) and inactive (negative) electrode and referenced to a common ground. (B) The main PVER electrode was placed along the midline at OZ. All active PVER electrodes were referenced to the inactive electrode at FZ. The common ground electrode was placed at CZ. (B) The PVER was also recorded above and below OZ using extra scalp electrodes placed at 5% inion- nasion distance above and below OZ.
The second difference was that the contrast of the checkerboard was varied over seven levels (0, 4, 9, 18, 37, 73, and 97%) to address the second hypothesis (H2B). The order of presentation was from low to high contrast to prevent uncontrolled carry over of adaptation effects. Likewise, to minimise adaptation, each checkerboard presentation was interleaved with a spatially homogenous grey stimulus (0% contrast) of the same average luminance (53 cd/m2) for one minute.
169 Responses to the checkerboard stimuli and blank screens were analysed in the same way, as per the methods in Chapter 2. With the exception of responses to 0% contrast, data were excluded if the 2F amplitude was less than noise (p > 0.05, or signal-to-noise ratio of at least 2.82), where noise was estimated from the average amplitude at the two neighbouring frequencies (14.6 and 18.8 Hz) (Meigen and Bach 1999). For responses at 0% contrast, the raw waveform and corresponding Fourier spectra consisted of random noise with no discernible response at the contrast reversal frequency. Thus, all 0% contrast data were analysed at 16.7 Hz and included in the final dataset as a means to infer the level of noise.
Amplitude and phase values were normalised to the maximum (97%) contrast condition to investigate relative changes in the cortical response, as per previous studies of human cortical contrast gain (Ross and Speed 1991; Truchard et al. 2000). To determine whether the response was monotonic or non-monotonic with
increasing contrast, a monotonicity index ( mi ) was calculated using Equation 4.3 (Ledgeway et al. 2005):
€ Rmax − R100 mi =1− (Equation 4.3) Rmax − R0
where Rmax was the maximum amplitude across all contrast levels tested, R100 was € the response at maximum contrast (97%), and R0 was the response at 0% contrast.
A monotonic response showed a value of mi = 1), whereas a non-monotonic, or € € supersaturating response, showed a value of mi < 1). € € 4.2.3.3 Modelling procedure €
Group data were modelled in Microsoft Excel (Microsoft, Redmond, WA, USA) to characterise the shape of the PVER contrast gain. Parameters were optimised using the Solver module by minimising the residual sum of squares ( RSS ), or merit function:
€
170
n ˆ 2 RSS = ∑(yi − y i ) (Equation 4.4)
th where yi was the i value of normalised amplitude ranging from 0 to 97% contrast € and yˆ i was the value of yi predicted by the model. During the optimisation process,
the parameters€ R0 , Rmax , c50, and n could either be fixed or floated. In all cases the € main parameters of interest, Rmax and c50, were allowed to float. However, as there € € were only seven levels of contrast (degrees of freedom = 6), floating all of the € € € € parameters yielded spurious and non-sensible solutions. Thus, to simplify the € € modelling procedure, R0 was fixed to the normalised amplitude at 0% contrast (Figure 4.7; controls = 0.07, migraine = 0.14), n was fixed to 1.4 from previous
literature (Busse et al. 2009; Tsai et al. 2011), whereas Rmax , c50, and s were floated. € Full details of the model comparisons that determined these constraints appear in € Appendix C, Section 9.2. € € €
4.2.4 Statistical analysis
4.2.4.1 Group comparisons
Statistical analysis was performed in SPSS version 20.0 (SPSS Inc., Chicago, IL, USA). Group comparisons were performed using a repeated-measures analysis of variance (RM-ANOVA) with a between-factor of group (control versus migraine) and a within-factor of contrast. Additionally, for the centre-surround data, the within-factor of contrast was further divided into the effects of the surround contrast and the centre contrast. The interaction effect was tested to address the two hypotheses of this experiment (H3A and H3B), as a significant interaction between group and contrast indicated that the difference between the two groups depended on stimulus contrast. Where the assumption of sphericity was violated according to
171 Mauchly’s test of sphericity (p > 0.05), the degrees of freedom were amended using a Huynh-Feldt correction.
4.2.4.2 Model comparisons
To establish whether there was a significant improvement (p < 0.05) in fitting the PVER data with a model consisting of more parameters, which decreased the degrees of freedom, an F-ratio was calculated:
(SS − SS )/SS F = 1 2 2 (Equation 4.5) (df1 − df 2 )/df2
where SS was the sum of squares and df was the degrees of freedom (number of € data points – number of parameters estimated by modelling) of the two models to be compared, where subscript 1 denoted the less complicated model (fewer parameters, € € more degrees of freedom) and subscript 2 denoted the more complicated model (more free parameters). F-ratios were then used to calculate p-values, where p < 0.05 indicated a significantly improved fit with the more complicated model.
4.3 Results
4.3.1 Perceptual centre-surround suppression
Figure 4.6 shows the group suppression ratio data. An increase in perceptual suppression was evident with increasing contrast of the surround (Figure 4.6, x-axis) for both control and migraine participants. Consistent with previous work (Ejima and Takahashi 1985; Xing and Heeger 2001), the amount of suppression was greatest when surround contrast was greater than centre contrast (Figure 4.6, data to the right of vertical dashed lines), with little suppression or facilitation when the contrast of
172 the centre was greater than that of the surround (Figure 4.6, data to the left of vertical dashed lines).
An RM-ANOVA revealed no overall difference in performance between groups (main effect of group: F (1, 42) = 1.41, p = 0.24). Neither was there a significant interaction between group and surround contrast (Huynh-Feldt ε = 0.69, F (2.07, 87.06) = 0.74, p = 0.82). However, there was a separation between groups in terms of perceptual suppression (Figure 4.6, shaded area) that was dependent on the contrast of the centre (group x centre interaction: Huynh-Feldt ε = 0.60, F (1.79, 75.31) = 4.37, p = 0.02).
The reduction in perceived contrast was most apparent for the 20% contrast centre condition (Figure 4.D), where people with migraine showed greater suppression with increasing surround contrast. On the other hand, for conditions that normally result in facilitation (Figure 4.6, unshaded area), the two groups performed the same. The maximum suppression ratio (i.e. facilitation) in the migraine group (SRmax = 1.08 ±
0.05) was comparable to that of control participants (SRmax = 1.08 ± 0.04). Thus, the imbalance between excitation and inhibition in the perceptual responses of people with migraine appeared to be driven by a change in inhibition, as facilitation (excitation) was normal but perceptual suppression was increased (inhibition).
The greatest amount of perceptual suppression in both control and migraine participants occurred when the stimulus consisted of a 20% contrast centre with a surround of 60% (controls: 0.73 ± 0.09, migraine: 0.68 ± 0.10) and 80% contrast (controls: 0.73 ± 0.11, migraine: 0.65 ± 0.13). Individual data for these conditions are shown in Figure 4.7. Although there is considerable overlap in perceptual performance between groups, a subset of individuals with migraine (17%) demonstrated significantly increased perceptual suppression, with suppression ratios below the minimum observed in the control group (SRmin = 0.56). These individuals have been depicted as cross symbols in subsequent figures.
173
Figure 4.6 – Mean suppression ratio (SR) as a function of surround contrast in the control (filled, n = 20) and migraine (unfilled, n = 24) groups for four centre contrast levels (Panels A to D). (A) 80% contrast centre (B) 60% contrast centre (C) 40% contrast centre (D) 20% contrast centre. The right panel of (C) plots the data from (Battista et al. 2011) (controls n = 20, migraine n = 26) for a similar drifting centre-surround stimulus of 95% contrast surround and 40% contrast centre. Error bars represent the 95% confidence interval of the mean. The horizontal dotted lines at 1 represents veridical perception. The vertical dashed lines indicate matched centre and surround contrast. SR < 1 indicates surround suppression (shaded area), whereas SR > 1 indicates facilitation (unshaded area). Group symbols have been jittered horizontally for clarity.
174
Figure 4.7 – Group mean and individual suppression ratios in the control (filled, n = 20) and migraine (unfilled, n = 24) groups when the centre contrast was 20% for two surround conditions where perceptual suppression (shaded area) was greatest: (A) 60% contrast surround (B) 80% contrast surround. Error bars represent the 95% confidence interval of the mean. The horizontal dotted lines at 1 represents veridical perception.
4.3.2 The pattern visual evoked response (cortical response)
4.3.2.1 PVER amplitude
To determine whether PVER amplitude was reduced relative to the control group across the range of contrast levels tested, amplitudes of both groups were normalised to the average control group amplitude at 97% contrast. These normalised data are plotted in Figure 4.8. There was no main effect of group (F (1, 37) = 0.54, p = 0.47). This result is not consistent with an overall reduction in the cortical response in the migraine group, as simulated in Figure 4.1. Neither was there a significant interaction between group and stimulus contrast (Huynh-Feldt ε = 0.30, F (1.78, 65.72) = 0.46, p = 0.61). Nevertheless, inspection of Figure 4.8 shows a trend for separation between groups at the low contrasts, with the average migraine cohort having a larger relative amplitude at low contrast that just failed to reach significance (note the error bars are 95% confidence intervals of the mean).
175
Figure 4.8 – Mean normalised PVER amplitude as a function of stimulus contrast in the control (filled, n = 20) and migraine (unfilled, n = 24) groups. Error bars represent the 95% confidence interval of the mean. Individual data were normalised to the average amplitude at 97% contrast of the control group.
The PVER data were also considered in terms of individual outcomes. Figure 4.9 illustrates a subset of the data at 0%, 18%, and 97% contrast. The 0% contrast data were unique because all data were included and were presumed to reflect non- significant random noise. At 0% contrast, people with migraine showed significantly higher amplitudes when analysed at the second harmonic (Figure 4.9A; controls: 0.06 ± 0.04, migraine: 0.11 ± 0.06; Mann Whitney rank sum test: p = 0.01), even though the 16.7 Hz data were non-significant according to Meigen and Bach (1999) relative to the neighbouring spectral values at 14.6 and 18.8 Hz (see Appendix C, Figure 9.2). Similarly, there was a trend for increased normalised PVER amplitude in the migraine group at 18% contrast (Figure 4.9B; controls: 0.71 ± 0.48, migraine: 0.82 ± 0.38; Mann Whitney rank sum test: p = 0.11). Both of these findings at low contrast are consistent with cortical hyperexcitability in migraine. At the highest contrast (97%), however, statistical comparisons showed no difference in PVER amplitude between the migraine and control groups (Figure 4.9C; controls: 1.00
176 ± 0.68, migraine: 1.01 ± 0.62; t42 = 0.05, p = 0.96), which conflicts with the earlier findings of this thesis. Possible explanations for the conflicting findings are discussed in the next chapter.
Figure 4.9 – Group median and individual normalised PVER amplitudes in the control (filled, n = 20) and migraine (unfilled, n = 24) groups at low and high contrasts. (A) 0% contrast (B) 18% contrast (C) 97% contrast. Error bars represent the interquartile range. Individual data were normalised to the average amplitude at 97% contrast of the control group. The four individuals from the migraine cohort who showed the most perceptual suppression from Figure 4.7 are shown as cross symbols.
4.3.2.2 PVER timing
The PVER timing data were also normalised to the average control group phase at 97% contrast. These are depicted in Figure 4.10. Phase values at each contrast level, and the change in phase with increasing contrast, were similar between groups (RM- ANOVA main effect of group: F (1, 37) = 0.07, p = 0.79; interaction: Huynh-Feldt ε = 0.35, F (2.11, 78.18) = 0.65, p = 0.53). During blank presentations of 0% contrast, phase assumed random values, similar to previous electrophysiological work in normal individuals (Porciatti and Ventura 2009a).
177
Figure 4.10 – Mean normalised PVER phase as a function of stimulus contrast in the control (filled, n = 20) and migraine (unfilled, n = 24) groups. Error bars represent the 95% confidence interval of the mean. Individual data were normalised to the average phase at 97% contrast in the control group.
Consistent with the steady-state PVER contrast gain of normal adults (Strasburger et al. 1993; Heinrich and Bach 2001; Garcia-Quispe et al. 2009), there was an initial positive slope (approximate 20% increase in phase) of the contrast response function, indicative of a signal delay in both control and migraine groups at low contrasts. After 25% contrast, phase remained approximately constant (Figure 4.10). At the maximum contrast (97%), phase values were not different between groups (controls 2.50 ± 0.25, migraine 2.52 ± 0.13). However, there was a trend in the migraine group to have more advanced signals (decreased phase) at lower contrasts (4% contrast: controls 2.16 ± 0.15, migraine 2.06 ± 0.28), which is also consistent with the increase in PVER amplitude, suggesting hyperexcitability at contrasts less than 25% (see Figure 4.8).
178 4.3.2.3 The shape of the PVER contrast gain
The simulations developed in Figure 4.2 indicate that if people with migraine were to express a hyperexcitable state, then the migraine group would show supersaturation with a characteristic ‘hump’ in the contrast gain. Inspection of the control group mean data in Figure 4.8 (filled circles) shows that the PVER amplitude increases monotonically across the full range of contrasts. The migraine group also showed a trend for increased amplitudes (Figure 4.8) and faster timing (Figure 4.10) at low contrasts (≤ 18%). This suggests that differences in PVER amplitude between groups might be contrast-dependent. The interaction between group and contrast was not, however, statistically significant (p = 0.61). Coupled with the large variation in individual outcomes (Figure 4.9) and the fact that supersaturation has been reported in a minority of primate cortical neurones (Ledgeway et al. 2005; Peirce 2007), it is possible that a real group trend was masked by having normalised the data to the average control group response. Hence, PVER amplitudes were normalised to the 97% contrast condition on an individual basis to describe the shape of an individual’s contrast response function. These data appear in Figure 4.11.
Modelling of the individually normalised data produced different outcomes in the control and migraine groups (Figure 4.11). For the control group, the saturating hyperbolic function (Equation 4.1) adequately described the data (Figure 4.11, solid curve; R2 = 0.97, F (1, 4) = 4.39, p = 0.10). This was not surprising given that, on average, the control participants showed a monotonic increase in PVER amplitude with contrast ( mi = 1, Equation 4.3). Representative raw waveforms and corresponding Fourier spectra of a control participant showing response saturation appears in Appendix C, Figure 9.2. On the other hand, the monotonicity index, mi , € of the mean migraine group response was 0.82, indicating that the PVER amplitude was not maximal at the highest contrast. Accordingly, the supersaturating function € (Equation 4.2) provided a significantly improved fit to these data (Figure 4.11, unfilled symbols, dashed curve; R2 = 0.98; F (1, 4) = 16.99 p = 0.02) with a suppressive exponent of 1.19 (95% CI: 0.89 – 1.72). This confirms that, on average,
179 the migraine group showed hyperexcitability at low contrast and supersaturation beyond approximately 20% contrast. Example raw waveforms and corresponding Fourier spectra of a migraine participant who showed response decline at high contrast is shown in Appendix C, Figure 9.2.
Figure 4.11 – Mean normalised PVER amplitude as a function of stimulus contrast in the control (filled, n = 20) and migraine (unfilled, n = 24) groups. Error bars represent the 95% confidence interval of the mean. Individual data were normalised to the amplitude at 97% contrast on an individual basis. The solid line is the saturating model fit (Equation 4.1) of the control group mean data (R2 = 0.97), whereas the dashed line is the supersaturating model fit (Equation 4.2) of the migraine group mean data (R2 = 0.98).
Monotonicity indices were also calculated for each individual as per Equation 4.3. Given that intra-sessional variability was approximately 10% for PVER amplitude measurements at 97% contrast (see Chapter 2), we considered a reduction in PVER amplitude of greater than 10%, or mi of less than 0.90, indicative of a significant response decline beyond that expected from measurement error. Three control participants and 10 migraine participants demonstrated a value of mi less than 0.90. € The difference between these proportions approached statistical significance at p <
€
180 0.05 (controls: 15%, migraine: 42%; chi-square: p = 0.05), indicating there were more individuals with migraine with an amplitude decline at high contrast.
As noted earlier, differences in PVER amplitude between migraine and non- headache control participants were not found based on overall group comparisons. Inspection of Figure 4.11 shows a substantial overlap of the error bars (95% confidence intervals of the mean) between the two groups, particularly at 18% contrast. At this contrast level, the variance found in the migraine group was significantly larger than for normal controls (F (23, 19) = 29.24, p < 0.001), indicating considerable variability in the shape of the contrast response function amongst migraine individuals.
Figure 4.12 provides an alternative way to visualise the difference in the shape of the group contrast response functions and also explains the larger error bars associated with the migraine group in Figure 4.11 by considering individual response characteristics. The majority of control amplitudes are maximal at 97% contrast, with a decreasing proportion of responses that are maximal at lower contrasts (Figure 4.12, filled bars). On the other hand, common with the perceptual test outcomes, there appear to be two trends in the migraine data. Half of the migraine sufferers showed normal characteristics (maximum amplitude at 97% contrast), whereas a subset of people showed varying degrees of hyperexcitability at low contrasts with supersaturation thereafter. Although the proportions at each contrast level were not significantly different between groups (Appendix C, Table 9.3; chi-square test of proportions: p > 0.05), Figure 4.12 shows a bi-modal distribution in the migraine group, with a tendency for more individuals with migraine than controls who show non-monotonic responses.
Thus, both the group and individual data support the alternative hypothesis put forth to explain the reduction in PVER amplitude in the previous chapters, i.e. that there is excessive excitatory feedback to inhibitory connections in migraine, resulting in a relative response decline at the highest contrasts (supersaturation).
181
Figure 4.12 – Histogram showing the proportions of control (filled, n = 20) and migraine (unfilled, n = 24) participants who showed maximum PVER amplitude in their individual contrast gain at the different contrast levels tested.
4.3.3 Relationships between measures
4.3.3.1 Relationship between perceptual and electrophysiological measures
In this experiment, perceptual and electrophysiological responses were obtained in the same individuals with migraine, both of which provided a measure of suppression or inhibition. The average suppression ratio from the two conditions that normally result in the greatest suppression (Figure 4.7; 20/60% and 20/80%) was used as a measure of the amount of perceptual suppression, whereas the monotonicity index ( mi , Equation 4.3) was used to infer the relative inhibitory effect on the PVER, or amount of supersaturation. Figure 4.13 depicts the relationship between these two measures. €
For the control participants, there was little correlation between the two measures (Spearman rank correlations: r = 0.01, p = 0.95), whereas the correlation coefficient for the migraine group approached statistical significance (Spearman r = 0.39, p =
182 0.06). This trend appears to be driven by a small number of individuals with migraine. Of the migraine participants who showed the most perceptual suppression (Figure 4.13, cross symbols), the two individuals with the most marked increase in suppression ratio were among the individuals who demonstrated the most attenuation of PVER amplitude at high contrast.
Figure 4.13 – Relationship between the amount of perceptual suppression and supersaturation of the PVER contrast response function in the control (filled, n = 20) and migraine (unfilled, n = 24) groups. The average suppression ratio across the two test conditions that normally invoke the greatest perceptual suppression (20% centre with
60% or 80% surround, see Figure 4.7) and the monotonicity index (mi ) were used as measures of suppression/inhibition. The four individuals from the migraine cohort who showed the most perceptual suppression from Figure 4.7 are shown as cross symbols. €
4.3.3.2 Relationship between visual function and migraine features
The two measures of suppression/inhibition obtained in this study (i.e. average suppression ratio, monotonicity index) were also used to determine whether visual
183 responses were related to any of the self-reported migraine characteristics described in Table 4.1. The table below shows that none of the correlation coefficients was significant for the entire migraine cohort (Table 4.2). A subset analysis of the data from the four migraine participants who showed the most perceptual suppression (see Figure 4.7) found no relationship between perceptual and electrophysiological measures and migraine characteristics.
Table 4.2 – Relationship between perceptual and electrophysiological measures of suppression/inhibition and self-reported characteristics of the migraine group (n = 24).
Suppression ratio Monotonicity index
a a Migraine characteristic R p-value R p-value
Days since last migraine 0.06 0.79 -0.08 0.73
Age at first migraine (years) 0.06 0.78 -0.21 0.33
Years of migraine -0.10 0.64 0.05 0.83
Weeks between migraines -0.20 0.36 -0.24 0.25
Estimated number of lifetime attacks -0.09 0.68 0.37 0.08
MIDAS questionnaire score (days) -0.04 0.87 -0.08 0.71 a p-values represent the significance of Spearman rank correlations
Table 4.3 – Relationship between perceptual and electrophysiological measures of suppression/inhibition and self-reported characteristics of the four participants who demonstrated the most perceptual suppression (see Figure 4.7).
Suppression ratio Monotonicity index
a a Migraine characteristic R p-value R p-value
Days since last migraine -0.20 0.92 -0.11 0.92
Age at first migraine (years) -0.80 0.33 -0.74 0.33
Years of migraine 0.40 0.75 -0.32 0.75
Weeks between migraines 0.21 0.92 -0.50 0.42
Estimated number of lifetime attacks 0.80 0.33 0.74 0.33
MIDAS questionnaire score (days) 0.20 0.92 0.11 0.92 a p-values represent the significance of Spearman rank correlations
184 4.4 Discussion
In this chapter, the following hypotheses were considered:
H3A: That perceptual surround suppression in people with migraine is increased for centre-surround stimuli of lower contrast centre and higher contrast surround.
This study builds on previous work (Battista et al. 2011) by measuring the perceived contrast for a range of different centre and surround contrast combinations. In support of the first hypothesis (H3A), there was a significant interaction between group and centre contrast (p = 0.02). In particular, the difference between control and migraine groups was most apparent for conditions where the centre contrast was low (Figure 4.6D), which were the conditions where suppression was normally elicited (Figure 4.6, shaded areas to the right of vertical dashed lines). Facilitation, on the other hand, was not affected (Figure 4.6, unshaded areas).
H3B: That the cortical electrophysiological response in people with migraine is increased for stimuli of lower contrast and decreased for stimuli of higher contrast.
For the first time in people with migraine, the PVER was recorded at a range of stimulus contrasts from 0 to 97%. Although there was no difference between control and migraine groups in PVER amplitude (Figure 4.8, p = 0.20) or timing (Figure 4.10; p = 0.79) across the different contrast levels, the migraine group showed a different pattern of PVER contrast gain (Figure 4.11). The control group data were better modelled using a saturating contrast response function (Figure 4.11, solid lines), indicating a monotonic increase in PVER amplitude with contrast ( mi = 1). On the other hand, the migraine group showed a significantly improved fit with Equation 4.2 (Figure 4.11, dashed lines; p = 0.02), which adequately reflected the € non-monotonic decline in PVER amplitude with increasing contrast ( mi = 0.82).
€ 185 PVER amplitudes tended to be higher in the migraine group at low contrasts (4%, 9%, 18%), which supports the theory of hyperexcitability at these contrasts. Furthermore, the data at 0% contrast was used to infer the level of noise. The increased amplitude at 16.7 Hz at 0% contrast in the migraine group can be interpreted as increased spontaneous neural activity. These results, taken together, provide support for the second hypothesis (H3B).
The next section discusses the general findings of this chapter in the context of the migraine literature.
4.4.1 Perceptual surround suppression in migraine
Previous psychophysical work, on which this study was based, identified relatively increased perceptual surround suppression in people with migraine compared to non-headache controls using a drifting contrast-contrast stimulus of 40% centre and 95% surround (Battista et al. 2011), which elicits a substantial shift in perceived contrast of approximately 20 to 40% reduction in normal observers. Like this study, Battista et al. (2011) found increased perceptual suppression in people with migraine, with no distinction between aura and non-aura groups. This provides convergent evidence for motion processing anomalies in both forms of migraine, consistent with studies using TMS (Battelli et al. 2002) and brain neuroimaging of area V5/MT (Antal et al. 2011) and psychophysical studies of motion processing (McKendrick and Badcock 2004c; Antal et al. 2005; Ditchfield et al. 2006; Shepherd et al. 2012).
This study adds a novel contribution to the literature by comparing perceptual suppression across a range of contrasts. The reduction in perceived contrast with increasing surround contrast, and the transition to facilitation when the centre contrast was higher than the surround, was observed in both migraine and control groups, consistent with previous studies (Ejima and Takahashi 1985; Xing and Heeger 2001; Yu et al. 2001). However, group differences were more prominent for low contrast stimuli, particularly when the centre contrast was low (Figure 4.6D). For
186 stimuli of high contrast (e.g. 80/80%), or with high contrast centres (Figure 4.6A), the surround had the same effect on both groups.
Neurophysiological studies at a single cell level provide possible insight as to why perceptual responses to high and low contrast conditions were different. Webb et al. (2005) showed that for low contrast centres, surround suppression was monocularly driven with broad spatiotemporal tuning. These characteristics lack the signature of most cortical neurones and are therefore attributed to an ‘early’, or pre-cortical, suppressive mechanism at the LGN or input layers of V1. Indeed, some surround suppression is already seen in the responses of LGN neurones (Solomon et al. 2002; Ozeki et al. 2004). On the other hand, surround suppression of high contrast centre stimuli is believed to originate from a ‘late’ cortical mechanism, which is binocularly driven, narrowly tuned to orientation and direction, and involves connections within and beyond V1 (Webb et al. 2005).
That our group differences were only significant for low contrast conditions does not necessarily imply a unique alteration in a pre-cortical suppressive mechanism in migraine. Rather, this result is likely to be a consequence of our monocular test protocol, which presumably did not invoke a strong suppressive response from binocularly driven cortical cells. To answer the question of whether pre-cortical or cortical mechanisms are likely to be contributing to the increased perceptual suppression in migraine requires further investigations comparing monocular (as per this study), binocular, and dichoptic stimulus presentations. A dichoptic experimental paradigm involves presentation of the centre stimulus to one eye, and presentation of the surround to the other eye, simultaneously. This also produces surround suppression, albeit the strength of suppression is weaker than when the centre and surround are presented to the same eye (DeAngelis et al. 1994; Petrov and McKee 2009).
Perceptual suppression was altered in the migraine group, but facilitation was normal. Of the several types of connections that are proposed to contribute to antagonistic centre-surround effects – excitatory feedforward and feedback
187 connections to V1 and intracortical inhibitory connections (Angelucci and Bressloff 2006; Schwabe et al. 2010) depicted in Figure 1.5 of Chapter 1 – increased intracortical inhibition is a possible candidate mechanism for the observations of this study, assuming that surround suppression occurred at the level of the cortex. Previous studies have hypothesised that an increase in the excitatory feedback to the lateral inhibitory connections in V1 could produce stronger suppressive effects of a surround in people with migraine (Battista et al. 2010; Battista et al. 2011).
An alternative explanation for stronger surround suppression is weakened adaptation to the surround. Normally, adaptation to the surround results in a decrease in its suppressive effect, i.e. an 80% surround effectively acts like a 60% surround (Cavanaugh et al. 2002; Durand et al. 2007) and generally occurs under longer stimulus presentations, although rapid contrast adaptation within as little as 30 ms also exists, as reviewed by Kohn (2007). In the present study, contrast-matching thresholds were measured using 500 ms stimulus presentation times, which was sufficient to produce significant adaptation effects (Kohn 2007). Future experiments could compare the effect of shorter presentation times to minimise the effect of adaptation (Karas 2012). If surround adaptation were altered in people with migraine, then the increased perceptual suppression observed in this study might not be present for shorter stimulus durations.
Although not directly comparable to the 95% contrast surround of the previous study by Battista et al. (2011), which produced an average suppression ratio of 0.53 in the migraine group (Figure 4.6C, right panel), the closest stimulus condition used in this study (40/80%) resulted in an average suppression ratio of 0.81 (Figure 4.6C, left panel). The smaller suppressive effect found in this study might be related to differences in the experimental paradigm. As mentioned earlier, perceptual responses were measured under monocular conditions, which may underestimate the suppressive effect from binocular sources. Whether binocular testing would produce a significantly greater suppressive effect, however, is uncertain. The binocular component of surround suppression in human observers is approximately three times weaker than the monocular component, yet dominates for stimulus durations
188 of more than 150 ms (Petrov and McKee 2009). In this study, the stimulus was presented for 500 ms to enable perception of motion at a drift rate of 2 degrees/second. Neither is it known whether a 95% contrast surround elicits significantly greater suppression compared to 80%, as the data of Figure 4.6D imply that the amount of perceptual suppression plateaus at approximately 60% surround contrast, and because studies of normal centre-surround interactions do not exceed test contrasts of 80% (Ejima and Takahashi 1985; Cannon and Fullenkamp 1991; Xing and Heeger 2001; Yu et al. 2001; Kilpelainen et al. 2007).
Another point of difference between the present study and that of Battista et al. (2011) was the introduction of a gap between the centre and surround (Figure 4.3), which is known to reduce the amount of perceptual suppression (Yu et al. 2001; Karas 2012). The gap, however, was necessary to facilitate the differentiation between the centre and surround under matched contrast conditions (Kilpelainen et al. 2007). Battista and colleagues also used a method of constant stimuli with seven contrast levels, each presented 20 times, to obtain psychometric functions comparing the perceived contrast of the ‘no-surround’ and the ‘surround’ conditions. The same MOCS procedure would have taken a considerably longer time to complete than the staircase adopted in this study. The number of trials used in this study is low compared to some laboratory-based psychophysics; however, that the two staircases starting above and below the veridical contrast converged at very similar contrast thresholds (Figure 4.4) provides an indication of the reliability of the results. Moreover, this study was designed to measure a range of perceptual and electrophysiological responses at the one test session on the same day, as migraine visual function is known to vary with time (see Chapter 3). Hence, the method of attaining contrast matching thresholds and the number of trials presented were chosen to enable measurement of a number of stimulus conditions (in this case, 16 different centre-surround combinations) within a realistic amount of time (no more than 2 hours) to avoid fatigue.
189 4.4.2 Cortical contrast gain in migraine
Studies of the PVER in migraine typically measure responses at a single, high contrast of at least 80% (for reviews, see Ambrosini et al. 2003 and Magis et al. 2007a), as was also the case for the previous experiments (Chapters 2 and 3). This is the first study to measure the PVER at a range of different contrasts to characterise the relative change in response, or contrast gain, by normalising the data to the maximum contrast level per individual. The normalised data were modelled with two different functions – a saturating and supersaturating model (Figure 4.11). On average, the control group showed a saturating response to contrast, similar to previous studies of normal human PVER (Morrone and Burr 1986; Bobak et al. 1988; Ross and Speed 1991; Heinrich and Bach 2001) and the majority of single cell responses in cat (Albrecht and Hamilton 1982; Ohzawa et al. 1985) and non-human primate visual cortex (Albrecht and Hamilton 1982; Sclar et al. 1989). On the other hand, the supersaturating function provided a better fit for the average migraine group data, accounting for the non-monotonic decline in amplitude at high contrasts. The approximate 18% decline can explain some of the reduction (approximately 33%) in steady-state PVER amplitude in both MO and MA groups at 97% contrast in the previous experiments of this thesis (Chapters 2 and 3). Furthermore, the supersaturating function adequately described the increase in amplitude at low contrasts (below 20%), which is here interpreted as evidence for hyperexcitability. This notion of hyperexcitability manifesting at low contrast is strengthened by the finding of significantly higher noise levels at 0% contrast in the migraine group, which has not previously been reported.
Throughout this thesis, the PVER has been used to infer the activity of visual cortical neurones. It must be remembered that the PVER also depends on feedback connections and the integrity of the preceding visual pathway (Fahle and Bach 2006). Monocular and dichoptic experimental paradigms in cat have demonstrated that the reduction in contrast gain of the cortical evoked visual response at mid to high contrast occurs primarily at a monocular site, before convergence of information
190 from the two eyes at a cortical level (Truchard et al. 2000). In humans, the PVER contrast response function is presumed to receive input from more than one retinogeniculate pathway (for a review, see Souza et al. 2012). Changes to the excitatory and inhibitory circuitry prior to V1 can therefore affect cortical contrast gain. For example, injection of GABA into the distinct magnocellular and parvocellular layers of the primate LGN attenuates the cortical response, although the magnitude of response reduction varies with the contrast-dependent dominance of each pathway (Allison et al. 2000). Thus, it is important to consider that differences in the supersaturating contrast gain of the migraine group might be attributed to abnormal cortical processing, but may also reflect reduced inputs from earlier, pre-cortical stages of the visual pathway. Supersaturation is seen in some retinal ganglion cells (Creutzfeldt et al. 1970) and neurones of the LGN (Kaplan et al. 1987). The level at which supersaturation takes place could not be determined in this study without pharmacological intervention or other invasive techniques. However, future studies could employ electrophysiological techniques to assess retinal function (e.g. PERG), as per the methods employed in Chapters 2 and 3.
Although the mechanism/s that shape the saturating and supersaturating contrast response functions are not completely understood, intracortical inhibition is believed to play an important role, assuming that the contrast gain observed in this study was cortical in origin. According to the ‘normalisation’ model (Heeger 1992; Carandini and Heeger 1994; Carandini et al. 1997), the cortical response is scaled (or ‘normalised’) with respect to contrast by the inhibitory activity of a ‘normalisation pool’ of cortical cells. A high contrast stimulus invokes stronger inhibition; hence, saturation normally occurs at high contrasts. In approximately 30% of primate V1 neurones, however, the response declines at high contrasts, which is presumed to arise from excessive lateral inhibition (Peirce 2007). This process is similar to the attenuation in human PVER by cross-orientation masking (Morrone and Burr 1986), which is greater at high than at low contrasts and parallels changes in GABA- mediated inhibitory processes observed in both single cell and PVER recordings in cat (Morrone et al. 1987). Hence, a mechanism of excessive lateral inhibition underlying the perceptual responses in our migraine group can be extended to
191 explain the supersaturating PVER contrast response function. This is consistent with more recent electrophysiological work demonstrating increased responses in people with migraine to a ‘windmill-dartboard’ stimulus (Coppola et al. 2013) – a central disc surrounded by three annuli that are radially divided into light and dark segments. Responses to temporal contrast reversal of this stimulus are believed to result from short-range lateral interactions (Zemon and Ratliff 1982).
An alternative explanation for changes in the PVER contrast gain in people with migraine is altered adaptation. PVER responses are suprathreshold responses, requiring contributions from a mass of neurones to evoke a significant surface potential. The checkerboard stimuli were presented for 750 ms before recording, as steady-state responses require at least repetitions of the stimulus before the steady- state level is attained (Armitage et al. 2001). Thereafter, the stimulus presentations were repeated, lasting approximately 20 seconds. As mentioned earlier in this discussion, contrast adaptation can occur in as little as 30 ms (Kohn 2007). Hence, the very nature of steady-state electrophysiological recording can produce effects of adaptation. Adaptation to flickering high-contrast patterns has recently been shown to be stronger in people with migraine, such that high contrasts were easier to discriminate, relative to control participants who were presumed to show minimal adaptation (Karanovic et al. 2011; Thabet et al. 2013). A possible explanation for this observation is that the adapted neural response function is steeper at the high contrast end, producing a much smaller increment threshold than the normal saturating response (Figure 4.14). Such a change in the saturating non-linearity of the visual system after adaptation has been modelled by Wilson and Humanski (1993) using feedback inhibition, which counteracts the effects of response saturation by recruitment of a network of neurones that reduce the magnitude of the contrast by division. This is similar in concept to the divisive normalisation model of Heeger (1992), which potentially explains the perceptual findings shown earlier and is compatible with the non-monotonic decline in PVER amplitude at high contrast observed in this study. The potential for adaptation effects are discussed more generally in the next chapter.
192
Figure 4.14 – Schematic illustrating how changes in the non-linearity of the PVER contrast gain might affect flicker contrast discrimination thresholds. The shape of the curves results in different contrast increments (∆C and ∆C’) for a given change in the neural response (∆R). A larger increment (∆C’) is required for the saturating contrast response function (solid line) compared to the contrast increment of the supersaturating function (∆C, dashed line) for the same magnitude of response (∆R). The corollary of this is that a fixed ∆C returns a smaller ∆R if the response saturates (solid line). The altered contrast gain might explain the improved flicker contrast discrimination thresholds observed in previous studies (Karanovic et al. 2011; Thabet et al. 2013). Adapted from Karanovic, O., Thabet. M., Wilson, H.R., and Wilkinson, F. (2011). “Detection and discrimination of flicker contrast in migraine.” Cephalalgia 31: 723 – 736.
To our knowledge, only two other studies have recorded the PVER in people with migraine at a level other than high contrast above 80%, which represents most electrophysiological studies in migraine (Ambrosini et al. 2003; Magis et al. 2007a). Zaletel et al. (2005) measured blood flow velocity and the transient PVER in response to checkerboards of 1%, 10%, and 100% contrast. Similar to the results of this experiment, Zaletel et al. (2005) did not find a significant difference in amplitude
193 across the contrasts tested. The PVER amplitude, however, was significantly higher in the MA group at 10% contrast, which is in agreement with the present trend of cortical hyperexcitability at similar low contrast levels (Figure 4.10).
Shibata et al. (2008) recorded the steady-state PVER and reported increased amplitudes in people with migraine at 29% and 98% contrast. The PVER amplitude was increased at low contrast, which is again similar to the study by Zaletel et al. (2005) and the present findings (Figure 4.10). It is not clear, however, whether there was a change in the shape of the migraine contrast response function. Normalising the group mean data from Shibata et al. (2008) at 29% contrast against 98% contrast demonstrates an approximate 20% reduction in amplitude relative to the maximal contrast condition. On the other hand, this study found an average 20% increase in normalised amplitude at about the same contrast (Figure 4.8). A point of difference between the present study and that by Shibata et al. (2008) is the order of stimulus presentation, which was not randomised. Cortical contrast gain is known to vary with the nature of sequential stimulation. Bonds (1991) showed that at a given contrast level, the response is greater when preceded by a lower contrast (ascending order) than when preceded by a higher contrast (descending order). We presented stimuli from low to high contrast, whereas Shibata and colleagues used a descending order. Thus, higher amplitudes might be expected with the experimental protocol of this study. This might explain the apparently conflicting results of relatively lower amplitudes observed by Shibata et al. (2008) and higher amplitudes observed in this study, when compared to the response at the highest contrast.
4.4.3 Relationship between perceptual and electrophysiological responses
The novel component of this study was the measurement of both perceptual and electrophysiological responses in the same individuals, to investigate whether changes in the migraine group were likely to be driven by excitatory or inhibitory dysfunction. The responses in the migraine group were altered in the same direction, consistent
194 with impaired inhibition. This implies a common underlying mechanism. One such mechanism is ‘divisive normalisation’ model of Heeger (1992). In this model, the ‘normalisation pool’ of cortical neurones that influence the primary cortical response is presumably activated by excitatory mechanisms (Heeger 1992), a process analogous to the excitatory feedback connections involved in centre-surround antagonism, which in turn excite inhibitory interneurones (Angelucci and Bressloff 2006; Schwabe et al. 2010). Hence, we expected to find a relationship between our measures of suppression/inhibition.
Although the correlation did not reach statistical significance at p < 0.05 (Figure 4.13, p = 0.06), there were individuals with migraine who displayed strong surround suppression as well as a significant decline in the PVER amplitude at high contrasts. Furthermore, the changes in the perceptual response to drifting centre-surround stimuli follow the trend predicted by the shape of the PVER contrast gain function. That is, increased amplitude and decreased phase at low contrast (supersaturation), consistent with cortical hyperexcitability. If altered visual responses in migraine reflect excessive excitation of inhibitory pathways, rather than a primary excitatory dysfunction that is consistent across all contrast levels, then a relative increase in the PVER at low contrasts (Figure 4.8) would invoke a relatively stronger inhibitory effect of the surround, with little or no change in inhibition at high contrasts. This can be seen in Figure 4.6 (vertical dashed lines), where the largest increase in perceptual suppression in people with migraine occurs at 20% matched contrast, which is approximately the level at which the relatively greatest electrophysiological response is elicited. Thus, both the perceptual and electrophysiological results are internally consistent with the idea of hyperexcitability in migraine. That the rank correlations failed to return significant outcomes possibly reflects the low statistical power resulting from only half of the migraine group showing non-monotonic responses (Figure 4.12).
Contrast-dependence of perceptual and electrophysiological abnormalities in migraine is potentially relevant to real-world functioning and the frequent observation that people with migraine find high-contrast patterns aversive (Marcus
195 and Soso 1989; Khalil 1991; Coleston and Kennard 1993; Shepherd 2000; Wilkinson and Crotogino 2000; Huang et al. 2003). The relative increase in inhibitory strength may be a compensatory mechanism to reduce the excessive neural activity elicited in response to a high contrast stimulus, which is similar to the ‘low pre-activation level’ hypothesised to counter the excessive excitability of the migraine cortex (Schoenen et al. 1995). Future work could include a measure of aversion, although highly subjective, to investigate the possible link with psychophysical and electrophysiological measures of visual function. The results of this study demonstrate that differences in migraine performance may or may not be present, depending on the specific contrast conditions used. It is recognised that variable results on electrophysiological testing in migraine are likely to be, in part, due to methodological differences (Ambrosini et al. 2003; Magis et al. 2007a). This study demonstrates that contrast is one stimulus variable that appears to influence the heterogeneity of PVER findings in migraine.
4.5 Conclusions
In summary, this study is the first demonstration that increased perceptual surround suppression and a relative reduction in PVER amplitude in migraine depend on the contrast of the stimulus. Not all individuals with migraine showed abnormal performance. Nevertheless, the general group trend for increased inhibition was evident in a subset of individuals (up to 50%), with some individuals showing markedly abnormal results on both perceptual and electrophysiological tests.
Both the perceptual and electrophysiological responses provide further support for an impaired excitatory/inhibitory balance in migraine. This imbalance manifests as a state of cortical hyperexcitability, as can be inferred from the increase and subsequent decrease in PVER amplitude (supersaturation) and the contrast- dependent increase in perceptual suppression. The proposed mechanism underlying these vision anomalies is an overall increase in the level of excitation in migraine that leads to excessive excitatory feedback and increased strength of lateral inhibition.
196 The two measures used in this study are non-invasive and have the potential to inform about differences in neural activity in the brain. The findings of this study can direct future work investigating which of the many connections and neural networks implicated in centre-surround processing and cortical evoked responses are possibly altered in migraine.
197
198 5 | Chapter Five General discussion and conclusions
5.1 Overview of findings
The major consistent finding throughout this thesis was that the cortical electrophysiological response to high contrast patterned stimuli (PVER) was abnormally reduced in the migraine-free interval. The series of experiments conducted in Chapters 2, 3, and 4 aimed to investigate possible reasons for this.
Firstly, the cortical deficit was not considered to be a downstream effect of retinal dysfunction, as the corresponding PERG was normal in Experiments 1 and 2 (Chapters 2 and 3). However, this does not preclude localised retinal involvement in migraine, as demonstrated in Experiment 2 (Chapter 3) by the small regions of visual field loss found in people with migraine. These localised visual field defects, sometimes monocular, were worse in the days immediately after migraine, as per previous work (McKendrick and Badcock 2004b), implying that migraine may have consequences outside of the brain. In Experiment 2 (Chapter 3), individuals with migraine showed visual field defects on different forms of perimetry – SAP, TMP, and SWAP – coinciding with other studies that demonstrate non-specific pre-cortical pathway dysfunction in migraine (Coleston et al. 1994; McKendrick and Badcock 2003; McKendrick and Sampson 2009).
Secondly, in Experiment 2 (Chapter 3), the cortical deficit remained stable in the days immediately after a migraine, despite further reductions in visual field sensitivity. These results suggest that the abnormal neural activity inferred from the PVER arises from a different mechanism than that underlying visual field defects. That is, rather
199 than a direct effect of migraine, the reduction in PVER might reflect a primary brain dysfunction that predisposes people to migraine. It is also possible that the reduction in PVER amplitude might reflect an epiphenomenon of migraine.
In the final experiment (Chapter 4), one of the competing theories behind cortical hyperexcitability in migraine was considered – a dynamic imbalance between excitation and inhibition (Vecchia and Pietrobon 2012). It was found that people with migraine showed altered contrast gain of both perceptual and electrophysiological responses, presumed to measure visual cortical function. At low contrasts, the net response was an increase in neural activity, whereas at high contrasts, there was a relatively greater effect of inhibition. Thus, cortical hyperexcitability in migraine cannot be thought of simplistically as a primary over- excitation or under-inhibition. Rather, the results provide support for an excitatory/inhibitory imbalance, which changes dynamically with the strength of stimulation, as a key candidate mechanism underlying cortical hyperexcitability in migraine.
A simplified schematic of how the findings of this thesis contribute to our current understanding of migraine pathophysiology is depicted in Figure 5.1. This schematic is based on a recent review paper, which proposed that migraine is a disorder of excitatory/inhibitory balance, and that sensory processing dysfunction, including vision anomalies in migraine, reflect this imbalance (Vecchia and Pietrobon 2012). Firstly, genetic models of familial hemiplegic migraine have provided some insight as to the possible nature of the primary brain dysfunction in migraine, i.e. excess glutamate and recurrent excitation. This leads to an interictal hyperexcitable brain state, with convergent evidence from psychophysical, electrophysiological, TMS, and brain neuroimaging studies (see Chapter 1). The abnormal resting state of hyperexcitability can trigger the initiation of CSD and activation of the trigeminovascular system, resulting in the symptoms of a migraine attack (aura, pain).
The decreased PVER amplitudes found throughout this thesis remained stable after a migraine event (Chapter 3) and were unlikely to be a downstream effect of retinal
200 dysfunction (Chapter 2). Rather, the decrease in the PVER as well as the increase in perceptual surround suppression in people with migraine observed in Experiment 3 (Chapter 4) could be explained by an excitatory/inhibitory imbalance in migraine, leading to increased inhibition via excessive contrast-dependent feedback. This is demonstrated by the abnormal contrast gain found in Experiment 3 (Chapter 4). The increase in PVER to 0% contrast found in Experiment 3 (Chapter 4) provides further support for cortical hyperexcitability in migraine, resulting in increased spontaneous neural activity, or noise.
Figure 5.1 – Proposed pathophysiological processes leading to vision anomalies in migraine. The features in black represent the current understanding of migraine pathophysiology, adapted from Vecchia, D. and Pietrobon, D. “Migraine: a disorder of brain excitatory-inhibitory balance?” Trends in Neuroscience 2012; 35: 507 – 520. The features in red represent the observations and hypotheses made in this thesis (see Chapter references). Processes that are still under debate in the migraine literature are depicted as dashed arrows. The black dashed box encompasses the processes that occur during the ictal period. Note that ‘homonymous visual field defects’ occur during migraine visual aura but might also persist in the interictal period.
201 5.2 Heterogeneity of findings
Based on group comparisons, the people with migraine showed reduced PVER amplitude, visual field deficits, and increased perceptual suppression. These findings, however, were not observed in every migraine sufferer, but was present, on average, in a larger subset of people with migraine than non-headache controls. A small group of non-headache controls showed similar visual anomalies as the migraine participants, although to a lesser degree. This brings to light the significant inter- individual variability in visual responses.
Even when comparing across the experiments of this thesis, the magnitude of the deficit found in people with migraine was not the same. As new people were recruited into each study, the heterogeneity of findings might be due to inter- individual variability between migraine observers, possibly owing to the variability in migraine characteristics, which can vary over a lifetime. Furthermore, because the experimental setup was not identical between Experiments 1 and 3 (Chapters 2 and 4), differences in the methodology must also be considered as a potential explanation for the variable findings in this thesis.
5.2.1 Differences in migraine characteristics
All participants with migraine in this thesis were diagnosed according to the International Headache Society classifications (International Headache Society 2013). Despite this, the MO and MA groups were not homogeneous in terms of their self- reported migraine characteristics, which might have contributed to the variable PVER findings. In Experiment 1 (Chapter 2), the MO participants reported greater disability scores on the MIDAS questionnaire. This is consistent with the higher attack frequency and generally more disabling nature of MO (International Headache Society 2013). On the other hand, the MA participants were significantly older and had suffered migraines for a longer period of time than MO participants. The MA group showed abnormalities (approximately 35% reduction) in the transient and
202 steady-state PVER, whereas deficits in the MO group (33% reduction) were only evident in the steady-state response. The steady-state response identifies greater dysfunction than transient responses in people with glaucoma (for a review, see Bach and Hoffmann 2008), in the same way as tests that bias processing to the M or K (i.e. less redundant) pre-cortical pathways have found defects in glaucoma earlier (Casson et al. 1993). Thus, one possible interpretation of the results of Experiment 1 (Chapter 2) is that shorter duration of migraine history in the MO participants resulted in less severe vision anomalies in these people, which did not manifest on recording the transient response.
Recruitment of milder cases of migraine might explain why the correlations between migraine history and attack frequency, suggestive of severity, and the magnitude of PVER defects were not significant throughout this thesis. Previous electrophysiological studies have recruited participants from hospitals or specialised headache and neurology clinics (e.g. Kennard et al. 1978; Boylu et al. 2010). These patients presumably represent the more severe cases of migraine (e.g. people with chronic or recalcitrant migraine or more severe attacks needing hospitalisation). In contrast, participants in the experiments of this thesis were recruited from the general population, primarily via advertisements placed at the University of Melbourne, with a minority of volunteers responding to an advertisement on the Headache Australia website. In addition, participants were excluded if they were taking regular prophylactic medications or suffered from chronic migraine. Thus, the participants of this thesis were arguably more representative of a broader clinical group of people with migraine.
The younger age of the participants in this thesis (up to 46 years old) compared to the study by Khalil et al. (2000), which recruited participants of up to 59 years of age, makes it possible that our participants suffered fewer total lifetime attacks. If, according to the suggestion by Kennard et al. (1978) and Khalil et al. (2000), that PVER deficits result from cumulative insult, then more severe or protracted cases would have been expected to show a larger magnitude of loss. However, paradoxically, the magnitude of the reduction in PVER found in this thesis (33%)
203 was greater than that of previous studies, which have reported an approximate 20% reduction in amplitude (Khalil et al. 2000; Boylu et al. 2010).
The non-significant correlations of this thesis should be viewed with caution, because correlating migraine characteristics with visual dysfunction was not the main purpose of the experiments. If this were the case, larger sample sizes would have been required to yield a high statistical power. Whether a larger study design would identify a relationship is uncertain, given that difficulties with retrospective reporting of migraine characteristics are well-recognised (see for example Russell et al. 1994; Wilkinson and Crotogino 2000; Coppola et al. 2007a). Self-reporting is subjective and relies on retrospection over years of migraine; thus, accurate estimates are limited or might be biased towards the most recent, more severe, or more interesting cases. A further complexity is that migraine features, such as attack frequency, can fluctuate over a lifetime, with the prevalence of migraine being highest between the ages of 25 and 55 years (Stewart et al. 1994; Bigal and Lipton 2009). The variable nature of migraines from one attack to the next also makes it difficult to generalise about the significance of a specific migraine characteristic. For example, the laterality of symptoms is not always conclusive, as the migraine aura or headache is not confined to the same side for every migraine attack (Coppola et al. 2007a) and approximately one-third of people with visual aura report bi-hemifield involvement (Russell and Olesen 1996).
5.2.2 Methodological differences
In Experiment 1 (Chapter 2), it was noted that a possible confound of previous work investigating the PVER in migraine was the influence of habituation, or lack thereof. Unlike the experiments in this thesis, many reports do not specify whether the PVER recordings were made continuously or with breaks in between stimulation. The deficient habituation observed in many studies across multiple sensory modalities (for a review, see Coppola et al. 2009) could increase the overall PVER amplitude, after averaging across all trials recorded under repetitive stimulation (Schoenen
204 1996). As breaks were implemented at regular intervals during recordings, we considered it unlikely that this effect would manifest in our results. Nevertheless, adaptation effects in electrophysiological recordings cannot be ruled out completely, because of the continuous nature of data collection and the need to average across many responses, at least over seconds, in order to attain a reasonable signal-to-noise.
In Experiment 3 (Chapter 4), the PVER amplitudes of the migraine group at the highest contrast were not significantly reduced relative to controls, due to some individuals with migraine showing very large amplitudes. It is possible that these larger amplitudes reflected an influence of adaptation as a consequence of measuring the PVER sequentially in response to increasing contrast, despite the introduction of a one-minute interval of grey screen between each contrast level. Deficient habituation, or adaptation, predicts that this would result in higher response amplitudes as a result of the preceding contrast level, particularly at higher contrasts. Data from two experiments of this thesis are compared in Figures 5.2A and 5.2B. On average, the means did not differ in Experiment 3 (Figure 5.2B). This is due to the very large amplitudes found in three people in the migraine group that were not observed in Experiment 1 (Figure 5.2A). Nevertheless, the addition of new data from Experiment 3 to the data of Experiment 1 did not change the conclusions of that study (Chapter 2) – that the migraine cohort as a whole showed a reduction in the steady-state PVER amplitude (Figure 5.2C; controls: n = 46, 3.44 ± 1.81, migraine: n
= 60, 2.51 ± 1.66; t104 = 2.74, p = 0.007).
The possibility that the different methodologies might have contributed to the discordant findings of Experiments 1 and 3 was further considered in a sub-analysis of participants who were tested in both experiments (Chapters 2 and 4). Data from the two eyes were averaged to enable paired comparisons, as illustrated in Figure 5.2D (control n = 4, migraine n = 9). Contrary to the control group (Figure 5.2D, left panel; paired t-tests: t3 = 0.02, p = 0.99), all of the migraine participants showed a significant increase in steady-state PVER amplitude (Figure 5.2D, right panel; paired t-tests: t8 = 4.07, p = 0.004). The increase in amplitude in the migraine group was consistent with a lack of habituation, as a result of sequential presentation of
205 increasing contrast levels. On the other hand, there was a trend for reduced responses in the control group (Figure 5.2D), which was in the direction predicted by normal habituation in this group.
Figure 5.2 – Group mean and individual PVER amplitudes in response to maximum contrast (97%) in the control (filled) and migraine (unfilled) groups in (A) Experiment 1 (controls: n = 30, migraine: n = 45) and (B) Experiment 3 (controls: n = 20, migraine: n= 24). (C) Data from new observers in Experiment 3 (cross symbols) were added to Experiment 1 to create a pooled dataset (controls: n = 46, migraine: n = 60). (D) Paired comparisons between Experiments 1 and 3 in the control (solid lines) and migraine (dashed lines) groups. Error bars represent the 95% confidence intervals of the mean.
206 Note that the above sub-analysis included only a small sample of observers. Nevertheless, Anderson and Vingrys (2001b) have argued that when serial success is found in a small sample, as is evident in the migraine group, then a sample size of five is adequate to interpret that observation as being representative of the population at large and not due to chance observation. Using this logic, it is possible that the experimental conditions of Experiment 3 (Chapter 4) might have produced a lack of habituation in the migraine cohort, with the magnitude varying among observers.
The other methodological difference was the use of extra recording electrodes above and below OZ in Experiment 3 (Chapter 4). This procedure was implemented in an attempt to minimise the inter-individual variability in cortical evoked responses due to differences in anatomy (Brindley 1972; Steinmetz et al. 1989; Hood and Zhang 2000; Ishikawa et al. 2011). Without neuroimaging measures to map the position of the visual cortex in the calcarine fissure relative to the location of electrode placement at OZ (Hood and Zhang 2000; Ishikawa et al. 2011), it was not possible in this thesis to verify that the difference in PVER measurements was not due to anatomical differences in people with migraine. Future investigations might address this issue by combining fMRI and PVER measures in the same individuals with migraine. Such a study design would also be useful to confirm that the PVER measures used in this thesis can be used as a surrogate marker for brain function in clinical practice, as the PVER is a relatively less expensive and more accessible tool to assay visual function than fMRI.
5.3 Possible non-visual mechanisms
It is possible that the differences in visual function found in this thesis are explicable by a non-visual mechanism. Investigations of cognitive function in people with migraine have yielded inconclusive results (for a review, see O'Bryant et al. 2005). Nevertheless, some studies have shown deficits in attention and memory in migraine sufferers (Mulder et al. 1999; Calandre et al. 2002), which could influence the visual
207 functional tests of this thesis. For example, the perception of contrast is enhanced with increased attention (Anton-Erxleben et al. 2010). Prolonged testing can induce drowsiness and an inability to maintain attention. Hence, a reduction in attention or development of fatigue might have contributed to the reduction in PVER amplitude found in response to checkerboard patterns in this thesis. Previous work, however, has indicated that dividing attention by performing a simultaneous auditory task (Mezer et al. 2004) or a mental task (e.g. counting numbers, see Uren et al. 1979) whilst viewing the PVER target does not significantly affect the waveform. In any case, participants were given regular breaks and encouraged to give verbal feedback throughout the session to confirm their attention. Likewise, a lack of attention to the central target on the perceptual centre-surround tasks might have reduced the perceived contrast of the stimulus in the migraine group. However, this would be expected to have produced an overall reduction in performance across all contrast levels, whereas the results of Experiment 3 (Chapter 4) clearly demonstrated a contrast-dependent alteration in the perceptual and electrophysiological responses. Furthermore, the localised visual field defects found in the second experiment (Chapter 3) are not consistent with the generalised changes expected if attention or fatigue were to have played a role.
Equally plausible is a lack of concentration due to heightened awareness of, or hypersensitivity to the visual stimuli (Marcus and Soso 1989; Coleston and Kennard 1993; Hay et al. 1994; Shepherd 2000; Huang et al. 2003; Kowacs et al. 2004). Participants in this thesis were not formally asked to rate their level of discomfort. Nevertheless, none of the participants needed to abort testing due to discomfort or aversion, and no one self-reported any discomfort. When migraine participants were contacted to obtain feedback about the test session and inquire if a migraine had occurred within three days of the test session, none of the participants recalled feeling discomfort immediately after the testing. Furthermore, the Plain Language Statement of each experiment explicitly described that the test sessions would involve the presentation of a range of visual stimuli, hence biasing the study sample to those without a known aversion to visual patterns. Previous studies have conducted tests of aversion, such as the flickering screen adjustment test of
208 (Wilkinson et al. 2008) or the visual discomfort questionnaire adopted by Shepherd et al. (2012). These tests are subjective and might introduce a bias in self-reporting, given that people with migraine show a tendency for hypochondria (Burker et al. 1989). However, criterion-free tests of aversion are yet to be developed and need to be implemented in future studies to test this possibility.
5.4 Potential usefulness of tests for clinical purposes
5.4.1 Differential diagnosis of migraine
The clinical diagnosis of migraine is a continuing challenge in practice. As migraine is diagnosed purely on medical history and symptomatology (International Headache Society 2013), it would be useful to have more objective measures to distinguish migraine headaches. Several investigators have previously considered the use of the PVER as a diagnostic test for migraine headache in children (Rossi et al. 1996; Lahat et al. 1997; Lahat et al. 1999; Bockowski et al. 2003), whose ability to describe their symptoms might be limited. Whilst not directly studied in this thesis, as all participants were adult migraine sufferers and non-headache controls, the overlap between the migraine and control groups and the significant inter-individual variability indicates that the visual tests used in this thesis are unlikely to be diagnostic clinical markers of the condition. Nevertheless, there were individuals with migraine in this thesis – recruited from a general population with otherwise normal vision and ocular health – who showed marked abnormalities in visual function outside the range of performance in controls.
5.4.2 Differential diagnosis of ocular disease
Visual symptoms often prompt patients to seek eye examinations. Visual field tests, in particular, are commonly conducted in clinical practice on indication when ocular and/or neurological disease is suspected. In this thesis, young migraine sufferers (19
209 to 46 years old) were largely unaware of the interictal visual field deficits found. These people might not otherwise have completed visual field tests in clinical practice, given their normal vision and ocular health and lack of migraine symptoms. Thus, it is important that clinicians are aware of the possibility for abnormal visual field test results in apparently healthy young individuals. In particular, the results of Experiment 2 (Chapter 3) confirmed previous findings that visual field defects are worse in the days immediately a migraine (Drummond and Anderson 1992; McKendrick and Badcock 2004b), which highlights the need to consider the timing of the test visit relative to the last migraine in interpretating visual field results.
The presence of visual field deficits in between migraine attacks has implications for the interpretation of test results in people who are suspected of ocular disease such as glaucoma. Some monocular and arcuate patterns of visual field loss in migraine have been described as ‘glaucomatous’ (Nizankowska et al. 1997; Çomoglu et al. 2003). The differential diagnosis of glaucoma is made more difficult with a concurrent history of migraine; hence, it might be prudent in these patients to enlist supplementary tests of visual function to help diagnose glaucomatous loss. For example, studies of the PERG in glaucoma patients have consistently demonstrated deficits in the retinal response in terms of timing and amplitude (Price et al. 1988; Bach and Speidel-Fiaux 1989; Pfeiffer and Bach 1992; Bach et al. 2006; Parisi et al. 2006; Tsaousis et al. 2013). All of the PERG parameters tested in this thesis, however, were normal despite abnormal visual field results. As glaucoma has been frequently associated with migraine in epidemiological studies (Corbett et al. 1985; Phelps and Corbett 1985; Orgul and Flammer 1994; Wang et al. 1997; Pradalier et al. 1998; Cursiefen et al. 2000; Hewitt et al. 2010), the possibility for concurrent disease was ruled out by ophthalmoscopic examination and imaging of the optic nerve head by the HRT. Intraocular pressures were also normal (below 21 mmHg). Previous studies have not reported the same level of detail regarding ocular health in their participants, if any (e.g. Boylu et al. 2010). The fact that the PERG was normal in this migraine cohort, following careful ocular examination, provides evidence that an abnormal PERG (in response to 97% contrast) is not characteristic of migraine sufferers.
210 In this thesis, abnormal visual function was identified in people with migraine by direct comparison with a group of non-headache controls. However, current visual field tests generate global indices to quantify dysfunction relative to an internal normative database, which does not exclude people with migraine. This can hamper the ability to detect an abnormal result in a migraine sufferer. Indeed, in Experiment 2 (Chapter 3), some individuals showed sensitivity losses compared with the non- headache control group, yet were not flagged as abnormal based on comparison with the manufacturer’s normative database. Likewise, laboratories are required to create a set of norms for electrophysiological tests (Odom et al. 2010; Bach et al. 2013). In this thesis, the migraine participants who were recruited suffered migraines regularly over the past year (but did not constitute a diagnosis of chronic migraine). Exclusion of such people from normative databases for PVER and visual field tests is therefore recommended to increase the likelihood of detecting abnormality using these clinical tests of vision.
5.5 Future work
This thesis has provided some insight as to the neural mechanism/s underlying specific vision anomalies in migraine. A number of further experiments have been identified to continue this line of research, as discussed next.
5.5.1 Anatomical locus of vision anomalies
The tests used in the final experiment (Chapter 4) were presumed to measure visual cortical function. Given that the PERG was consistently normal throughout this thesis, the responses observed on PVER measurement were considered to reflect cortical activity. In Experiment 3 (Chapter 4), it was important as a first step to conduct the study monocularly to enable comparison with the previous PVER findings of this thesis. However, monocular presentation of the centre-surround stimuli in Experiment 3 (Chapter 4) was not sufficient to pinpoint the origin of the
211 perceptual response. Given that both monocular and binocular mechanisms contribute to surround suppression (Petrov et al. 2005; Webb et al. 2005), further insight could be gained by a comparison between monoptic and dichoptic paradigms. The differences in surround suppression in Experiment 3 (Chapter 4), where the stimuli were presented monocularly, could arise at pre-cortical and cortical stages of the visual pathway. However, if differences were only apparent by presenting the centre stimulus to one eye, and the surround stimulus to the other eye (dichoptic stimulation), it is possible to confirm that the abnormality is occurring after the level of binocular combination, i.e. at a cortical locus. Such an approach has been used more recently by Thabet et al. (2013) to ascertain that the critical flicker adaptation processes that contributes to altered flicker contrast processing in migraine occurs pre-cortically, possibly at the level of the LGN or within the retina.
Alternatively, the orientation of the stimulus surround relative to the centre also influences the amount of suppression elicited. When the centre and surround consist of orthogonal grating patterns, perceptual suppression is reduced (Xing and Heeger 2001; Yu et al. 2001), with parallel findings in animal studies when V1 neurones are stimulated with the same stimuli (Levitt and Lund 1997; Cavanaugh et al. 2002). Webb et al. (2005) measured the spatiotemporal tuning of V1 neurones in primate to low and high centre contrasts. The authors demonstrated that centre-surround responses to low contrast showed broad orientation tuning and were monocularly driven. On the other hand, surround suppression at V1 was tightly tuned for orientation and driven binocularly. A future study comparing whether differences in perceptual suppression between groups are broadly or tightly tuned to orientation would also assist in pinpointing the likely anatomical locus for surround suppressive effects in migraine.
212 5.5.2 Time course of changes in migraine
5.5.2.1 Vision after migraine
The second experiment of this thesis (Chapter 3) was motivated by the observation of atypical resolution of interictal visual field defects over time. In some cases, visual field deficits have resolved within weeks or months (Ebner 1991; Sullivan-Mee and Bowman 1997), whereas others persist and can even be permanent (Ebner 1991; McKendrick et al. 2000; Razeghinejad et al. 2009). As only two test sessions were conducted in this experiment, it was not possible to determine the natural progression of visual field losses in migraine. Longitudinal data on the long-term significance of visual field deficits in people with migraine would be useful for at least two reasons: to determine whether the presence of visual field deficits in the interictal period might be predictive of future eye disease, and to help establish guidelines as to the monitoring of migraine patients over time. Such a long-term study design is fraught with difficulty, however, as migraine is an episodic condition and susceptibility to migraine attacks appears to fluctuate over a lifetime. Moreover, the longevity of visual field deficits varies considerably from individual to individual. Nevertheless, it might be possible to track changes in visual function over a person’s migraine cycle from one attack to the next on a case-by-case basis. Knowing the progression of visual field changes over time could potentially inform management of patients using effective migraine therapy to minimise ocular damage in those individuals who show losses of sensitivity after a migraine event.
5.5.2.2 Vision in between migraines
Tracking visual function daily over a person’s migraine cycle is also potentially useful for the treatment of migraine. Knowing the time-course of changes in the status of an individual may indicate whether a migraine is likely to occur. This can, in turn, provide guidelines for the effective use of prophylactic anti-migraine medications. Previous work using electrophysiology and advanced neuroimaging has identified
213 changes in brain excitability up to 72 hours leading up to a migraine event (Judit et al. 2000; Sand and Vingen 2000; Sand et al. 2008; Chen et al. 2009; Sand et al. 2009; Siniatchkin et al. 2009). Such measures of visual function on a daily basis would be a logistical nightmare. Instead, measures of visual function should ideally be quick and easy to perform, and possibly be conducted without the need for supervision or laboratory equipment. Psychophysical tests could satisfy these requirements, particularly with the advent of portable devices, as these tests need only access to a computer (with response buttons) and screen.
One such psychophysical test could be the perceptual analogue of centre-surround antagonism used in Experiment 3 (Chapter 4). At present, it is not known whether perceptual surround suppression changes with time, as the experiments of this thesis and others (Battista et al. 2010; Battista et al. 2011) have measured responses at a single time point during the interictal period. Presuming that perceptual suppression reflects an underlying neural dysfunction responsible for migraine susceptibility, and that neural dysfunction is progressively intensified leading up to a migraine (Vecchia and Pietrobon 2012), differences in perceptual suppression might be most apparent in the pre-attack period. In that case, the amount of perceptual suppression could therefore alert a person of, and enable preparation for an impending attack. It could also provide a metric for assessing efficacy of prophylactic medications for migraine.
5.5.3 Physiological correlates for behavioural anomalies
Human performance on visual tasks such as the centre-surround task necessarily involves pooling of a large number of neural responses. Similarly, the VER is also a massed response from the visual cortex (Fahle and Bach 2006). Both of these responses are governed by excitatory and inhibitory processes, and alterations in either or both of these processes could be responsible for the abnormal neural state of the migraine brain. Despite convergent evidence from a range of visual functional measures, few studies other than this thesis have considered concurrent psychophysical and electrophysiological anomalies in the same cohort of migraine
214 participants (Khalil 1991). The abnormality of inhibitory function found in Experiment 3 (Chapter 4) was in the same direction for both perceptual and electrophysiological tasks, which argues for a common pathophysiological basis that is likely generalisable to many other sensory tests.
Supplementary brain imaging techniques might be potentially useful to confirm areas that are responsible for the deficits observed on psychophysical and electrophysiological testing. For example, schizophrenia has been hypothesised to result from reduced GABA concentration and neurotransmission. Yoon et al. (2010) demonstrated that in vivo measurements using magnetic resonance spectroscopy of reduced GABA levels in the visual cortex of 13 schizophrenic patients was highly correlated with reduced perceptual surround suppression. This provides further insight into the neural mechanism/s of cognitive deficits in schizophrenia as involving an abnormality in GABA-ergic processing. It remains to be seen whether increased perceptual surround suppression in people with migraine observed in this thesis and previous studies (Battista et al. 2010; Battista et al. 2011) is related to dynamic changes in GABA levels with contrast, as would be predicted from the altered contrast gain found in Experiment 3 (Chapter 4).
5.6 Conclusions
People with migraine show vision anomalies in between attacks, even though asymptomatic, which has implications for tests of visual function in clinical practice and research settings. The findings of this thesis demonstrate that not one mechanism can adequately explain all of the vision anomalies that are found interictally. In addition to cortical hyperexcitability, the predominant neural anomaly in migraine, there is the potential for vision anomalies that reflect pre-cortical dysfunction and/or adverse sequelae of migraine. This thesis also demonstrates the utility of non-invasive tests of visual function as surrogate markers of a generalised sensory processing dysfunction in migraine. However, given the inter-individual variability in responses, it is not always useful to only consider the group differences.
215 Rather, the susceptibility to migraine and the risk of adverse effects of migraine may need to be considered individually, so that treatment regimens can be implemented on a case-by-case basis.
216 6 | References
Abdul-Rahman, A.M., Gilhotra, J.S. and Selva, D. (2011). "Dynamic focal retinal arteriolar vasospasm in migraine." Indian Journal of Ophthalmology 59: 51-53.
Afra, J., Cecchini, A.P., De Pasqua, V., Albert, A. and Schoenen, J. (1998). "Visual evoked potentials during long periods of pattern-reversal stimulation in migraine." Brain 121: 233-241.
Afridi, S.K., Giffin, N.J., Kaube, H., Friston, K.J., Ward, N.S., Frackowiak, R.S. and Goadsby, P.J. (2005). "A positron emission tomographic study in spontaneous migraine." Archives of Neurology 62: 1270-1275.
Airy, H. (1870). "On a distinct form of transient hemiopsia." Philosophical Transactions of the Royal Society of London 160: 247-270.
Akerman, S., Holland, P.R. and Goadsby, P.J. (2011). "Diencephalic and brainstem mechanisms in migraine." Nature Reviews. Neuroscience 12: 570-584.
Albrecht, D.G. and Hamilton, D.B. (1982). "Striate cortex of monkey and cat: contrast response function." Journal of Neurophysiology 48: 217-237.
Albrecht, D.G., Farrar, S.B. and Hamilton, D.B. (1984). "Spatial contrast adaptation characteristics of neurones recorded in the cat's visual cortex." The Journal of Physiology 347: 713-739.
Allison, J.D., Melzer, P., Ding, Y., Bonds, A.B. and Casagrande, V.A. (2000). "Differential contributions of magnocellular and parvocellular pathways to the contrast response of neurons in bush baby primary visual cortex (V1)." Visual Neuroscience 17: 71-76.
Allman, J., Miezin, F. and McGuinness, E. (1985). "Direction- and velocity-specific responses from beyond the classical receptive field in the middle temporal visual area (MT)." Perception 14: 105-126.
Aloisi, P., Marrelli, A., Porto, C., Tozzi, E. and Cerone, G. (1997). "Visual evoked potentials and serum magnesium levels in juvenile migraine patients." Headache 37: 383-385.
Alvarez, W.C. (1960). "The migrainous scotoma as studied in 618 persons." American Journal of Ophthalmology 49: 489-504.
Ambrosini, A., de Noordhout, A.M., Sandor, P.S. and Schoenen, J. (2003). "Electrophysiological studies in migraine: a comprehensive review of their interest and limitations." Cephalalgia 23: 13-31.
American Clinical Neurophysiology Society (2006). "Guideline 5: Guidelines for standard electrode position nomenclature." Journal of Clinical Neurophysiology 23: 107-110.
Anderson, A.J. and Vingrys, A.J. (2000). "Interactions between flicker thresholds and luminance pedestals." Vision Research 40: 2579-2588.
Anderson, A.J. and Vingrys, A.J. (2001a). "Multiple processes mediate flicker sensitivity." Vision Research 41: 2449-2455.
217 Anderson, A.J. and Vingrys, A.J. (2001b). "Small samples: does size matter?" Investigative Ophthalmology and Visual Science 42: 1411-1413.
Anderson, D. (1992). Automated Static Perimetry. St Louis, Missouri, Mosby.
Anderson, R.S., Redmond, T., McDowell, D.R., Breslin, K.M. and Zlatkova, M.B. (2009). "The robustness of various forms of perimetry to different levels of induced intraocular stray light." Investigative Ophthalmology and Visual Science 50: 4022-4028.
Angelucci, A. and Bressloff, P.C. (2006). "Contribution of feedforward, lateral and feedback connections to the classical receptive field center and extra-classical receptive field surround of primate V1 neurons." Progress in Brain Research 154: 93-120.
Antal, A., Temme, J., Nitsche, M.A., Varga, E.T., Lang, N. and Paulus, W. (2005). "Altered motion perception in migraineurs: evidence for interictal cortical hyperexcitability." Cephalalgia 25: 788-794.
Antal, A., Polania, R., Saller, K., Morawetz, C., Schmidt-Samoa, C., Baudewig, J., Paulus, W. and Dechent, P. (2011). "Differential activation of the middle-temporal complex to visual stimulation in migraineurs." Cephalalgia 31: 338-345.
Anton-Erxleben, K., Abrams, J. and Carrasco, M. (2010). "Evaluating comparative and equality judgments in contrast perception: attention alters appearance." Journal of Vision 10(11): 6.
Armitage, J.A., Bui, B.V., Gibson, R. and Vingrys, A.J. (2001). "Postnatal development of flicker sensitivity in guinea pigs." Clinical and Experimental Optometry 84: 270-275.
Artes, P.H., Iwase, A., Ohno, Y., Kitazawa, Y. and Chauhan, B.C. (2002). "Properties of perimetric threshold estimates from Full Threshold, SITA Standard, and SITA Fast strategies." Investigative Ophthalmology and Visual Science 43: 2654-2659.
Arthurs, O.J., Williams, E.J., Carpenter, T.A., Pickard, J.D. and Boniface, S.J. (2000). "Linear coupling between functional magnetic resonance imaging and evoked potential amplitude in human somatosensory cortex." Neuroscience 101: 803-806.
Aurora, S.K., Ahmad, B.K., Welch, K.M., Bhardhwaj, P. and Ramadan, N.M. (1998). "Transcranial magnetic stimulation confirms hyperexcitability of occipital cortex in migraine." Neurology 50: 1111- 1114.
Aurora, S.K. and Wilkinson, F. (2007). "The brain is hyperexcitable in migraine." Cephalalgia 27: 1442- 1453.
Bach, M., Hiss, P. and Rover, J. (1988). "Check-size specific changes of pattern electroretinogram in patients with early open-angle glaucoma." Documenta Ophthalmologica 69: 315-322.
Bach, M. and Speidel-Fiaux, A. (1989). "Pattern electroretinogram in glaucoma and ocular hypertension." Documenta Ophthalmologica 73: 173-181.
Bach, M., Pfeiffer, N. and Birkner-Binder, D. (1992). "Pattern-electroretinogram reflects diffuse retinal damage in early glaucoma." Clinical Vision Sciences 7: 335-340.
Bach, M. and Meigen, T. (1999). "Do's and don'ts in Fourier analysis of steady-state potentials." Documenta Ophthalmologica 99: 69-82.
Bach, M. and Mathieu, M. (2004). "Different effect of dioptric defocus vs. light scatter on the pattern electroretinogram (PERG)." Documenta Ophthalmologica 108: 99-106.
218 Bach, M., Unsoeld, A.S., Philippin, H., Staubach, F., Maier, P., Walter, H.S., Bomer, T.G. and Funk, J. (2006). "Pattern ERG as an early glaucoma indicator in ocular hypertension: a long-term, prospective study." Investigative Ophthalmology and Visual Science 47: 4881-4887.
Bach, M. and Hoffmann, M.B. (2008). "Update on the pattern electroretinogram in glaucoma." Optometry and Vision Science 85: 386-395.
Bach, M., Brigell, M.G., Hawlina, M., Holder, G.E., Johnson, M.A., McCulloch, D.L., Meigen, T. and Viswanathan, S. (2013). "ISCEV standard for clinical pattern electroretinography (PERG): 2012 update." Documenta Ophthalmologica 126: 1-7.
Bahra, A., Matharu, M., Buchel, C., Frackowiak, R. and Goadsby, P. (2001). "Brainstem activation specific to migraine headache." Lancet 357: 1016-1017.
Baker, C.L., Jr., Hess, R.F. and Zihl, J. (1991). "Residual motion perception in a "motion-blind" patient, assessed with limited-lifetime random dot stimuli." The Journal of Neuroscience 11: 454-461.
Barbanti, P., Fofi, L., Aurilia, C. and Egeo, G. (2013). "Dopaminergic symptoms in migraine." Neurological Sciences 34: S67-70.
Barlow, H.B. (1956). "Retinal noise and absolute threshold." Journal of the Optical Society of America 46: 634-639.
Barlow, H.B. and Hill, R.M. (1963). "Evidence for a physiological explanation of the waterfall phenomenon and figural after-effects." Nature 200: 1345-1347.
Bartel, P., Becker, P. and Robinson, E. (1991). "The intrasession repeatability of pattern electroretinograms and the effects of digital filtering." Documenta Ophthalmologica 76: 351-358.
Battelli, L., Black, K.R. and Wray, S.H. (2002). "Transcranial magnetic stimulation of visual area V5 in migraine." Neurology 58: 1066-1069.
Battista, J., Badcock, D.R. and McKendrick, A.M. (2010). "Center-surround visual motion processing in migraine." Investigative Ophthalmology and Visual Science 51: 6070-6076.
Battista, J., Badcock, D.R. and McKendrick, A.M. (2011). "Migraine increases centre-surround suppression for drifting visual stimuli." PLoS One 6: e18211.
Benedek, K., Tajti, J., Janaky, M., Vecsei, L. and Benedek, G. (2002). "Spatial contrast sensitivity of migraine patients without aura." Cephalalgia 22: 142-145.
Bengtsson, B. and Heijl, A. (2006). "Diagnostic sensitivity of fast blue-yellow and standard automated perimetry in early glaucoma: a comparison between different test programs." Ophthalmology 113: 1092- 1097.
Berninger, T. (1986). "The pattern electroretinogram and its contamination." Clinical Vision Sciences 1: 185-190.
Beversdorf, D., Stommel, E., Allen, C., Stevens, R. and Lessell, S. (1997). "Recurrent branch retinal infarcts in association with migraine." Headache 37: 396-399.
Bigal, M.E. and Lipton, R.B. (2009). "The epidemiology, burden, and comorbidities of migraine." Neurologic Clinics 27: 321-334.
Blau, J.N. (1980). "Migraine prodromes separated from the aura: complete migraine." British Medical Journal 281: 658-660.
219 Blau, J.N. (1991). "Migraine postdromes: symptoms after attacks." Cephalalgia 11: 229-231.
Bobak, P., Bodis-Wollner, I. and Marx, M.S. (1988). "Cortical contrast gain control in human spatial vision." The Journal of Physiology 405: 421-437.
Bockowski, L., Sobaniec, W., Smigielska-Kuzia, J., Kulak, W. and Solowiej, E. (2003). "[The pattern- reversal visual evoked potentials in children with migraine with aura and without aura]." Roczniki Akademii Medycznej w Bialymstoku 48: 154-157.
Bohotin, V., Fumal, A., Vandenheede, M., Gerard, P., Bohotin, C., Maertens de Noordhout, A. and Schoenen, J. (2002). "Effects of repetitive transcranial magnetic stimulation on visual evoked potentials in migraine." Brain 125: 912-922.
Bohotin, V., Fumal, A., Vandenheede, M., Bohotin, C. and Schoenen, J. (2003). "Excitability of visual V1-V2 and motor cortices to single transcranial magnetic stimuli in migraine: a reappraisal using a figure-of-eight coil." Cephalalgia 23: 264-270.
Bolay, H., Reuter, U., Dunn, A.K., Huang, Z., Boas, D.A. and Moskowitz, M.A. (2002). "Intrinsic brain activity triggers trigeminal meningeal afferents in a migraine model." Nature Medicine 8: 136-142.
Bonds, A.B. (1991). "Temporal dynamics of contrast gain in single cells of the cat striate cortex." Visual Neuroscience 6: 239-255.
Bowd, C., Tafreshi, A., Vizzeri, G., Zangwill, L.M., Sample, P.A. and Weinreb, R.N. (2009). "Repeatability of pattern electroretinogram measurements using a new paradigm optimized for glaucoma detection." Journal of Glaucoma 18: 437-442.
Bowerman, L. (1989). "Transient visual field loss secondary to migraine." Journal of the American Optometric Association 60: 912-916.
Boylu, E., Domac, F.M., Kocer, A., Unal, Z., Tanridag, T. and Us, O. (2010). "Visual evoked potential abnormalities in migraine patients." Electromyography and Clinical Neurophysiology 50: 303-308.
Brigell, M., Bach, M., Barber, C., Moskowitz, A. and Robson, J. (2003). "Guidelines for calibration of stimulus and recording parameters used in clinical electrophysiology of vision." Documenta Ophthalmologica 107: 185-193.
Brighina, F., Palermo, A. and Fierro, B. (2009). "Cortical inhibition and habituation to evoked potentials: relevance for pathophysiology of migraine." The Journal of Headache and Pain 10: 77-84.
Brigo, F., Storti, M., Tezzon, F., Manganotti, P. and Nardone, R. (2013). "Primary visual cortex excitability in migraine: a systematic review with meta-analysis." Neurological Sciences 34: 819-830.
Brinciotti, M., Guidetti, V., Matricardi, M. and Cortesi, F. (1986). "Responsiveness of the visual system in childhood migraine studied by means of VEPs." Cephalalgia 6: 183-185.
Brindley, G.S. (1972). "The variability of the human striate cortex." The Journal of Physiology 225: 1-3.
Broadway, D.C. and Drance, S.M. (1998). "Glaucoma and vasospasm." British Journal of Ophthalmology 82: 862-870.
Burker, E.J., Hannay, J. and Halsey, J. (1989). "Neuropsychological functioning and personality characteristics of migrainous and nonmigrainous female college students." Neuropsychology 3: 61-73.
Burr, D.C. and Morrone, M.C. (1987). "Inhibitory interactions in the human vision system revealed in pattern-evoked potentials." The Journal of Physiology 389: 1-21.
220 Busse, L., Wade, A.R. and Carandini, M. (2009). "Representation of concurrent stimuli by population activity in visual cortex." Neuron 64: 931-942.
Calandre, E.P., Bembibre, J., Arnedo, M.L. and Becerra, D. (2002). "Cognitive disturbances and regional cerebral blood flow abnormalities in migraine patients: their relationship with the clinical manifestations of the illness." Cephalalgia 22: 291-302.
Callaway, E.M. (2005). "Structure and function of parallel pathways in the primate early visual system." The Journal of Physiology 566: 13-19.
Cannon, M.W. and Fullenkamp, S.C. (1991). "Spatial interactions in apparent contrast: inhibitory effects among grating patterns of different spatial frequencies, spatial positions and orientations." Vision Research 31: 1985-1998.
Cao, Y., Welch, K.M., Aurora, S. and Vikingstad, E.M. (1999). "Functional MRI-BOLD of visually triggered headache in patients with migraine." Archives of Neurology 56: 548-554.
Cao, Y., Aurora, S.K., Nagesh, V., Patel, S.C. and Welch, K.M. (2002). "Functional MRI-BOLD of brainstem structures during visually triggered migraine." Neurology 59: 72-78.
Carandini, M. and Heeger, D.J. (1994). "Summation and division by neurons in primate visual cortex." Science 264: 1333-1336.
Carandini, M., Heeger, D.J. and Movshon, J.A. (1997). "Linearity and normalization in simple cells of the macaque primary visual cortex." The Journal of Neuroscience 17: 8621-8644.
Carandini, M. and Ferster, D. (1997). "A tonic hyperpolarization underlying contrast adaptation in cat visual cortex." Science 276: 949-952.
Carandini, M., Movshon, J.A. and Ferster, D. (1998). "Pattern adaptation and cross-orientation interactions in the primary visual cortex." Neuropharmacology 37: 501-511.
Carter, O.L., Pettigrew, J.D., Hasler, F., Wallis, G.M., Liu, G.B., Hell, D. and Vollenweider, F.X. (2005). "Modulating the rate and rhythmicity of perceptual rivalry alternations with the mixed 5-HT2A and 5-HT1A agonist psilocybin." Neuropsychopharmacology 30: 1154-1162.
Casagrande, V.A., Yazar, F., Jones, K.D. and Ding, Y. (2007). "The morphology of the koniocellular axon pathway in the macaque monkey." Cerebral Cortex 17: 2334-2345.
Casson, E.J., Johnson, C.A. and Shapiro, L.R. (1993). "Longitudinal comparison of temporal- modulation perimetry with white-on-white and blue-on-yellow perimetry in ocular hypertension and early glaucoma." Journal of the Optical Society of America A. Optics, Image Science, and Vision 10: 1792-1806.
Cavanaugh, J.R., Bair, W. and Movshon, J.A. (2002). "Nature and interaction of signals from the receptive field center and surround in macaque V1 neurons." Journal of Neurophysiology 88: 2530-2546.
Celesia, G.G. and Kaufman, D. (1985). "Pattern ERGs and visual evoked potentials in maculopathies and optic nerve diseases." Investigative Ophthalmology and Visual Science 26: 726-735.
Celesia, G.G., Kaufman, D. and Cone, S. (1987). "Effects of age and sex on pattern electroretinograms and visual evoked potentials." Electroencephalography and Clinical Neurophysiology 68: 161-171.
Chatterjee, S. and Callaway, E.M. (2003). "Parallel colour-opponent pathways to primary visual cortex." Nature 426: 668-671.
221 Chen, W.T., Wang, S.J., Fuh, J.L., Lin, C.P., Ko, Y.C. and Lin, Y.Y. (2009). "Peri-ictal normalization of visual cortex excitability in migraine: an MEG study." Cephalalgia 29: 1202-1211.
Chronicle, E. and Mulleners, W. (1994). "Might migraine damage the brain?" Cephalalgia 14: 415-418.
Chubb, C., Sperling, G. and Solomon, J.A. (1989). "Texture interactions determine perceived contrast." Proceedings of the National Academy of Sciences of the United States of America 86: 9631-9635.
Cohen, J. (1988). Statistical Power Analysis for the Behavioural Sciences. Hillsdale, NJ, Lawrence Earlbaum Associates.
Coleston, D.M. and Kennard, C. (1993). "Visual changes in migraine: indications of cortical dysfunction." Cephalalgia 13: 11.
Coleston, D.M., Chronicle, E., Ruddock, K.H. and Kennard, C. (1994). "Precortical dysfunction of spatial and temporal visual processing in migraine." Journal of Neurology, Neurosurgery, and Psychiatry 57: 1208-1211.
Coleston, D.M. and Kennard, C. (1995). "Responses to temporal visual stimuli in migraine: the critical flicker fusion test." Cephalalgia 15: 396-398.
Cologno, D., Torelli, P. and Manzoni, G.C. (1998). "Migraine with aura: a review of 81 patients at 10- 20 years' follow-up." Cephalalgia 18: 690-696.
Çomoglu, S., Yarangümeli, A., Köz, O.G., Elhan, A.H. and Kural, G. (2003). "Glaucomatous visual field defects in patients with migraine." Journal of Neurology 250: 201-206.
Connolly, J.F., Gawel, M. and Rose, F.C. (1982). "Migraine patients exhibit abnormalities in the visual evoked potential." Journal of Neurology, Neurosurgery, and Psychiatry 45: 464-467.
Coppeto, J.R., Lessell, S., Sciarra, R. and Bear, L. (1986). "Vascular retinopathy in migraine." Neurology 36: 267-270.
Coppola, G., Parisi, V., Fiermonte, G., Restuccia, R. and Pierelli, F. (2007a). "Asymmetric distribution of visual evoked potentials in patients with migraine with aura during the interictal phase." European Journal of Ophthalmology 17: 828-835.
Coppola, G., Ambrosini, A., Di Clemente, L., Magis, D., Fumal, A., Gerard, P., Pierelli, F. and Schoenen, J. (2007b). "Interictal abnormalities of gamma band activity in visual evoked responses in migraine: an indication of thalamocortical dysrhythmia?" Cephalalgia 27: 1360-1367.
Coppola, G., Pierelli, F. and Schoenen, J. (2007c). "Is the cerebral cortex hyperexcitable or hyperresponsive in migraine?" Cephalalgia 27: 1427-1439.
Coppola, G., Pierelli, F. and Schoenen, J. (2009). "Habituation and migraine." Neurobiology of Learning and Memory 92: 249-259.
Coppola, G., Curra, A., Sava, S.L., Alibardi, A., Parisi, V., Pierelli, F. and Schoenen, J. (2010a). "Changes in visual-evoked potential habituation induced by hyperventilation in migraine." The Journal of Headache and Pain 11: 497-503.
Coppola, G., Curra, A., Serrao, M., Di Lorenzo, C., Gorini, M., Porretta, E., Alibardi, A., Parisi, V. and Pierelli, F. (2010b). "Lack of cold pressor test-induced effect on visual-evoked potentials in migraine." The Journal of Headache and Pain 11: 115-121.
222 Coppola, G., Cremers, J., Gerard, P., Pierelli, F. and Schoenen, J. (2011). "Effects of light deprivation on visual evoked potentials in migraine without aura." BMC Neurology 11: 91.
Coppola, G., Parisi, V., Di Lorenzo, C., Serrao, M., Magis, D., Schoenen, J. and Pierelli, F. (2013). "Lateral inhibition in visual cortex of migraine patients between attacks." The Journal of Headache and Pain 14: 20-31.
Corbett, J.J., Phelps, C.D., Eslinger, P. and Montague, P.R. (1985). "The neurologic evaluation of patients with low-tension glaucoma." Investigative Ophthalmology and Visual Science 26: 1101-1104.
Creutzfeldt, O.D., Sakmann, B., Scheich, H. and Korn, A. (1970). "Sensitivity distribution and spatial summation within receptive-field center of retinal on-center ganglion cells and transfer function of the retina." Journal of Neurophysiology 33: 654-671.
Cursiefen, C., Wisse, M., Cursiefen, S., Junemann, A., Martus, P. and Korth, M. (2000). "Migraine and tension headache in high-pressure and normal-pressure glaucoma." American Journal of Ophthalmology 129: 102-104.
Dacey, D.M. (1996). "Circuitry for color coding in the primate retina." Proceedings of the National Academy of Sciences of the United States of America 93: 582-588.
Dacey, D.M., Peterson, B.B., Robinson, F.R. and Gamlin, P.D. (2003). "Fireworks in the primate retina: in vitro photodynamics reveals diverse LGN-projecting ganglion cell types." Neuron 37: 15-27.
Dakin, S., Carlin, P. and Hemsley, D. (2005). "Weak suppression of visual context in chronic schizophrenia." Current Biology 15: 822-824.
Dalkara, T., Zervas, N.T. and Moskowitz, M.A. (2006). "From spreading depression to the trigeminovascular system." Neurological Sciences 27: S86-90.
De Natale, R.D., Polimeni, D., Narbone, M.C., Scullica, M.G. and Pellicano, M. (1993). Visual field defects in migraine patients. In: Mills RP (ed). Perimetry Update 1992/93. Proceedings of the Xth International Perimetric Society Meeting. Amsterdam/New York, Kugler Publications: pp 283-284.
DeAngelis, G.C., Freeman, R.D. and Ohzawa, I. (1994). "Length and width tuning of neurons in the cat's primary visual cortex." Journal of Neurophysiology 71: 347-374.
Demirel, S. and Vingrys, A.J. (1994). "Eye movements during perimetry and the effect that fixational instability has on perimetric outcomes." Journal of Glaucoma 3: 28-35.
Denuelle, M., Fabre, N., Payoux, P., Chollet, F. and Geraud, G. (2008). "Posterior cerebral hypoperfusion in migraine without aura." Cephalalgia 28: 856-862.
Derrington, A.M. and Henning, G.B. (1981). "Pattern discrimination with flickering stimuli." Vision Research 21: 597-602.
Derrington, A.M. and Lennie, P. (1984). "Spatial and temporal contrast sensitivities of neurones in lateral geniculate nucleus of macaque." The Journal of Physiology 357: 219-240.
Dersu, II, Thostenson, J., Jane Durcan, F., Hamilton, S.M. and Digre, K.B. (2012). "Optic disc and visual test findings in patients with migraine." Journal of Clinical Neuroscience 20: 72-74.
Di Clemente, L., Coppola, G., Magis, D., Fumal, A., De Pasqua, V. and Schoenen, J. (2005). "Nociceptive blink reflex and visual evoked potential habituations are correlated in migraine." Headache 45: 1388-1393.
223 Di Russo, F., Pitzalis, S., Spitoni, G., Aprile, T., Patria, F., Spinelli, D. and Hillyard, S.A. (2005). "Identification of the neural sources of the pattern-reversal VEP." Neuroimage 24: 874-886.
Di Russo, F., Pitzalis, S., Aprile, T., Spitoni, G., Patria, F., Stella, A., Spinelli, D. and Hillyard, S.A. (2007). "Spatiotemporal analysis of the cortical sources of the steady-state visual evoked potential." Human Brain Mapping 28: 323-334.
Diener, H.C., Scholz, E., Dichgans, J., Gerber, W.D., Jack, A., Bille, A. and Niederberger, U. (1989). "Central effects of drugs used in migraine prophylaxis evaluated by visual evoked potentials." Annals of Neurology 25: 125-130.
Ditchfield, J.A., McKendrick, A.M. and Badcock, D.R. (2006). "Processing of global form and motion in migraineurs." Vision Research 46: 141-148.
Douthwaite, W.A. and Connor, H. (1989). "Mental concentration and the pattern reversal visual evoked response." Optometry and Vision Science 66: 61-65.
Drance, S., Anderson, D.R. and Schulzer, M. (2001). "Risk factors for progression of visual field abnormalities in normal-tension glaucoma." American Journal of Ophthalmology 131: 699-708.
Drummond, P.D. (1987). "Pupil diameter in migraine and tension headache." Journal of Neurology, Neurosurgery, and Psychiatry 50: 228-230.
Drummond, P.D. and Anderson, M. (1992). "Visual field loss after attacks of migraine with aura." Cephalalgia 12: 349-352.
Ducros, A., Denier, C., Joutel, A., Cecillon, M., Lescoat, C., Vahedi, K., Darcel, F., Vicaut, E., Bousser, M.G. and Tournier-Lasserve, E. (2001). "The clinical spectrum of familial hemiplegic migraine associated with mutations in a neuronal calcium channel." The New England Journal of Medicine 345: 17-24.
Dumoulin, S.O., Bittar, R.G., Kabani, N.J., Baker, C.L., Jr., Le Goualher, G., Bruce Pike, G. and Evans, A.C. (2000). "A new anatomical landmark for reliable identification of human area V5/MT: a quantitative analysis of sulcal patterning." Cerebral Cortex 10: 454-463.
Durand, S., Freeman, T.C. and Carandini, M. (2007). "Temporal properties of surround suppression in cat primary visual cortex." Visual Neuroscience 24: 679-690.
Dustman, R.E. and Beck, E.C. (1969). "The effects of maturation and aging on the wave form of visually evoked potentials." Electroencephalography and Clinical Neurophysiology 26: 2-11.
Ebner, R. (1991). "Visual field examination during transient migrainous visual loss." Journal of Clinical Neuroophthalmology 11: 114-117.
Eccles, J.C. (1951). "Interpretation of action potentials evoked in the cerebral cortex." Electroencephalography and Clinical Neurophysiology 3: 449-464.
Efron, B. and Gong, G. (1983). "A leisurely look at the bootstrap, the jackknife and cross-validation." The American Statistician 37: 36-48.
Eisner, A., Austin, D.F. and Samples, J.R. (2004). "Short wavelength automated perimetry and tamoxifen use." British Journal of Ophthalmology 88: 125-130.
Ejima, Y. and Takahashi, S. (1985). "Apparent contrast of a sinusoidal grating in the simultaneous presence of peripheral gratings." Vision Research 25: 1223-1232.
224 Etminan, M., Takkouche, B., Isorna, F.C. and Samii, A. (2005). "Risk of ischaemic stroke in people with migraine: systematic review and meta-analysis of observational studies." British Medical Journal 330: 63-67.
Fahle, M. and Bach, M. (2006). Origin of the visual evoked potentials. In: Heckenlively JR and Arden GB (eds). Principles and Practice of Clinical Electrophysiology of Vision. Cambridge, Massachusetts, MIT Press: pp 207-234.
Ferrari, M.D., Haan, J., Blokland, J.A., Arndt, J.W., Minnee, P., Zwinderman, A.H., Pauwels, E.K. and Saxena, P.R. (1995). "Cerebral blood flow during migraine attacks without aura and effect of sumatriptan." Archives of Neurology 52: 135-139.
Fierro, B., Ricci, R., Piazza, A., Scalia, S., Giglia, G., Vitello, G. and Brighina, F. (2003). "1 Hz rTMS enhances extrastriate cortex activity in migraine: evidence of a reduced inhibition?" Neurology 61: 1446- 1448.
Flammer, J., Pache, M. and Resink, T. (2001). "Vasospasm, its role in the pathogenesis of diseases with particular reference to the eye." Progress in Retinal and Eye Research 20: 319-349.
Flammer, J., Konieczka, K. and Flammer, A.J. (2013). "The primary vascular dysregulation syndrome: implications for eye diseases." The European Association for Predictive Preventive and Personalised Medicine Journal 4: 14-47.
Froehlich, J. and Kaufman, D.I. (1991). "Effect of decreased retinal illumination on simultaneously recorded pattern electroretinograms and visual-evoked potentials." Investigative Ophthalmology and Visual Science 32: 310-318.
Fumal, A., Coppola, G., Bohotin, V., Gerardy, P.Y., Seidel, L., Donneau, A.F., Vandenheede, M., Maertens de Noordhout, A. and Schoenen, J. (2006). "Induction of long-lasting changes of visual cortex excitability by five daily sessions of repetitive transcranial magnetic stimulation (rTMS) in healthy volunteers and migraine patients." Cephalalgia 26: 143-149.
Galezowski, X. (1882). "Ophthalmic megrim: An affection of the vasomotor nerves of the retina and retinal centre which may end in a thrombosis." Lancet 119: 176-177.
Gallant, J.L., Connor, C.E., Rakshit, S., Lewis, J.W. and Van Essen, D.C. (1996). "Neural responses to polar, hyperbolic, and Cartesian gratings in area V4 of the macaque monkey." Journal of Neurophysiology 76: 2718-2739.
Gallant, J.L., Shoup, R.E. and Mazer, J.A. (2000). "A human extrastriate area functionally homologous to macaque V4." Neuron 27: 227-235.
Garcia-Quispe, L.A., Gordon, J. and Zemon, V. (2009). "Development of contrast mechanisms in humans: a VEP study." Optometry and Vision Science 86: 708-716.
Gardner, J.L., Sun, P., Waggoner, R.A., Ueno, K., Tanaka, K. and Cheng, K. (2005). "Contrast adaptation and representation in human early visual cortex." Neuron 47: 607-620.
Gasser, P. and Flammer, J. (1987). "Influence of vasospasm on visual function." Documenta Ophthalmologica 66: 3-18.
Gasser, P., Flammer, J., Guthauser, U. and Mahler, F. (1990). "Do vasospasms provoke ocular diseases?" Angiology 41: 213-220.
Gasser, P. and Meienberg, O. (1991). "Finger microcirculation in classical migraine. A video- microscopic study of nailfold capillaries." European Neurology 31: 168-171.
225 Gipponi, S., Scaroni, N., Venturelli, E., Forbice, E., Rao, R., Liberini, P., Padovani, A. and Semeraro, F. (2013). "Reduction in retinal nerve fiber layer thickness in migraine patients." Neurological Sciences 34: 841-845.
Glass, L. (1969). "Moire effect from random dots." Nature 223: 578-580.
Glenn, A.M., Shaw, P.J., Howe, J.W. and Bates, D. (1992). "Complicated migraine resulting in blindness due to bilateral retinal infarction." British Journal of Ophthalmology 76: 189-190.
Goadsby, P.J. and Edvinsson, L. (1993). "The trigeminovascular system and migraine: studies characterizing cerebrovascular and neuropeptide changes seen in humans and cats." Annals of Neurology 33: 48-56.
Goodale, M.A. and Milner, A.D. (1992). "Separate visual pathways for perception and action." Trends in Neurosciences 15: 20-25.
Goodwin, D. (2011). "Transient complete homonymous hemianopia associated with migraine." Optometry 82: 298-305.
Granziera, C., DaSilva, A.F., Snyder, J., Tuch, D.S. and Hadjikhani, N. (2006). "Anatomical alterations of the visual motion processing network in migraine with and without aura." PLoS Medicine 3: e402.
Graveson, G.S. (1949). "Retinal arterial occlusion in migraine." British Medical Journal 2: 838-840.
Greenlee, M.W. (2000). "Human cortical areas underlying the perception of optic flow: brain imaging studies." International Review of Neurobiology 44: 269-292.
Grüsser, O. (1995). "Migraine phosphenes and the retino-cortical magnification factor." Vision Research 35: 1125-1134.
Gustavsson, A., Svensson, M., Jacobi, F., Allgulander, C., Alonso, J., Beghi, E., Dodel, R., Ekman, M., Faravelli, C., Fratiglioni, L., Gannon, B., Jones, D.H., Jennum, P., Jordanova, A., Jonsson, L., Karampampa, K., Knapp, M., Kobelt, G., Kurth, T., Lieb, R., Linde, M., Ljungcrantz, C., Maercker, A., Melin, B., Moscarelli, M., Musayev, A., Norwood, F., Preisig, M., Pugliatti, M., Rehm, J., Salvador- Carulla, L., Schlehofer, B., Simon, R., Steinhausen, H.C., Stovner, L.J., Vallat, J.M., Van den Bergh, P., van Os, J., Vos, P., Xu, W., Wittchen, H.U., Jonsson, B. and Olesen, J. (2011). "Cost of disorders of the brain in Europe 2010." European Neuropsychopharmacology 21: 718-779.
Gutteridge, I.F., McDonald, R.A. and Plenderleith, J.G. (2007). "Branch retinal artery occlusion during a migraine attack." Clinical and Experimental Optometry 90: 371-375.
Hadjikhani, N., Sanchez Del Rio, M., Wu, O., Schwartz, D., Bakker, D., Fischl, B., Kwong, K.K., Cutrer, F.M., Rosen, B.R., Tootell, R.B., Sorensen, A.G. and Moskowitz, M.A. (2001). "Mechanisms of migraine aura revealed by functional MRI in human visual cortex." Proceedings of the National Academy of Sciences of the United States of America 98: 4687-4692.
Hardage, L. and Tyler, C.W. (1995). "Induced twinkle aftereffect as a probe of dynamic visual processing mechanisms." Vision Research 35: 757-766.
Harle, D.E. and Evans, B.J. (2005). "Frequency doubling technology perimetry and standard automated perimetry in migraine." Ophthalmic and Physiological Optics 25: 233-239.
Harrison, J.M., O'Connor, P.S., Young, R.S., Kincaid, M. and Bentley, R. (1987). "The pattern ERG in man following surgical resection of the optic nerve." Investigative Ophthalmology and Visual Science 28: 492-499.
226 Harrison, W.W., Viswanathan, S. and Malinovsky, V.E. (2006). "Multifocal pattern electroretinogram: cellular origins and clinical implications." Optometry and Vision Science 83: 473-485.
Hay, K.M., Mortimer, M.J., Barker, D.C., Debney, L.M. and Good, P.A. (1994). "1044 women with migraine: the effect of environmental stimuli." Headache 34: 166-168.
Heeger, D.J. (1992). "Normalization of cell responses in cat striate cortex." Visual Neuroscience 9: 181- 197.
Heeger, D.J. and Ress, D. (2002). "What does fMRI tell us about neuronal activity?" Nature Reviews. Neuroscience 3: 142-151.
Hegyalijai, T., Meienberg, O., Dubler, B. and Gasser, P. (1997). "Cold-induced acral vasospasm in migraine as assessed by nailfold video-microscopy: prevalence and response to migraine prophylaxis." Angiology 48: 345-349.
Heijl, A. and Krakau, C.E. (1977). "A note on fixation during perimetry." Acta Ophthalmologica 55: 854- 861.
Heine, M. and Meigen, T. (2004). "The dependency of simultaneously recorded retinal and cortical potentials on temporal frequency." Documenta Ophthalmologica 108: 1-8.
Heinrich, T.S. and Bach, M. (2001). "Contrast adaptation in human retina and cortex." Investigative Ophthalmology and Visual Science 42: 2721-2727.
Hendrickson, A.E., Wilson, J.R. and Ogren, M.P. (1978). "The neuroanatomical organization of pathways between the dorsal lateral geniculate nucleus and visual cortex in Old World and New World primates." The Journal of Comparative Neurology 182: 123-136.
Hendry, S. and Reid, R. (2000). "The koniocellular pathway in primate vision." Annual Review of Neuroscience 23: 127-153.
Hendry, S.H. and Yoshioka, T. (1994). "A neurochemically distinct third channel in the macaque dorsal lateral geniculate nucleus." Science 264: 575-577.
Heravian-Shandiz, J., Douthwaite, W.A. and Jenkins, T.C. (1992). "Effect of attention on the VEP in binocular and monocular conditions." Ophthalmic and Physiological Optics 12: 437-442.
Hering, R. and Kuritzky, A. (1992). "Sodium valproate in the prophylactic treatment of migraine: a double-blind study versus placebo." Cephalalgia 12: 81-84.
Hess, R.F. and Baker Jr, C.L. (1984). "Human pattern-evoked electroretinogram." Journal of Neurophysiology 51: 939-951.
Hewitt, A., Wu, J., Green, C., Lai, T., Kearns, L., Craig, J. and Mackey, D. (2010). "Systemic disease associations of familial and sporadic glaucoma: the Glaucoma Inheritance Study in Tasmania." Acta Ophthalmologica 88: 70-74.
Hill, D.L., Daroff, R.B., Ducros, A., Newman, N.J. and Biousse, V. (2007). "Most cases labeled as "retinal migraine" are not migraine." Journal of Neuroophthalmology 27: 3-8.
Ho, T.W., Edvinsson, L. and Goadsby, P.J. (2010). "CGRP and its receptors provide new insights into migraine pathophysiology." Nature Reviews. Neurology 6: 573-582.
227 Hoeffken, O., Stude, P., Lenz, M., Bach, M., Dinse, H.R. and Tegenthoff, M. (2009). "Visual paired- pulse stimulation reveals enhanced visual cortex excitability in migraineurs." European Journal of Neuroscience 30: 714-720.
Holder, G.E. (1989). "Pattern electroretinography in patients with delayed pattern visual evoked potentials due to distal anterior visual pathway dysfunction." Journal of Neurology, Neurosurgery, and Psychiatry 52: 1364-1368.
Holder, G.E. (1991). "The incidence of abnormal pattern electroretinography in optic nerve demyelination." Electroencephalography and Clinical Neurophysiology 78: 18-26.
Holliday, I.E. and Ruddock, K.H. (1983). "Two spatio-temporal filters in human vision. 1. Temporal and spatial frequency response characteristics." Biological Cybernetics 47: 173-190.
Hood, D.C. and Zhang, X. (2000). "Multifocal ERG and VEP responses and visual fields: comparing disease-related changes." Documenta Ophthalmologica 100: 115-137.
Hood, D.C., Xu, L., Thienprasiddhi, P., Greenstein, V.C., Odel, J.G., Grippo, T.M., Liebmann, J.M. and Ritch, R. (2005). "The pattern electroretinogram in glaucoma patients with confirmed visual field deficits." Investigative Ophthalmology and Visual Science 46: 2411-2418.
Hood, D.C., Bach, M., Brigell, M., Keating, D., Kondo, M., Lyons, J.S., Marmor, M.F., McCulloch, D.L. and Palmowski-Wolfe, A.M. (2012). "ISCEV standard for clinical multifocal electroretinography (mfERG) (2011 edition)." Documenta Ophthalmologica 124: 1-13.
Huang, J., Cooper, T.G., Satana, B., Kaufman, D.I. and Cao, Y. (2003). "Visual distortion provoked by a stimulus in migraine associated with hyperneuronal activity." Headache 43: 664-671.
Huang, J., DeLano, M. and Cao, Y. (2006). "Visual cortical inhibitory function in migraine is not generally impaired: evidence from a combined psychophysical test with an fMRI study." Cephalalgia 26: 554-560.
Hubel, D.H. and Wiesel, T.N. (1968). "Receptive fields and functional architecture of monkey striate cortex." The Journal of Physiology 195: 215-243.
Hupp, S.L., Kline, L.B. and Corbett, J.J. (1989). "Visual disturbances of migraine." Survey of Ophthalmology 33: 221-236.
Hykin, P.G., Gartry, D., Brazier, D.J. and Graham, E. (1991). "Bilateral cilio-retinal artery occlusion in classic migraine." Postgraduate Medical Journal 67: 282-284.
Iester, M., Mikelberg, F.S., Swindale, N.V. and Drance, S.M. (1997). "ROC analysis of Heidelberg Retina Tomograph optic disc shape measures in glaucoma." Canadian Journal of Ophthalmology 32: 382- 388.
Inan, L.E., Uysal, H., Ergun, U., Yurdakul, M. and Karagoz, H. (1994). "Complicated retinal migraine." Headache 34: 50-52.
International Headache Society (2004). "The International Classification of Headache Disorders: 2nd edition." Cephalalgia 24: S9-160.
International Headache Society (2013). "The international classification of headache disorders (3rd edition)." Cephalalgia 33: 629-808.
228 Ishikawa, K., Nagai, T., Yamada, Y., Negi, A. and Nakamura, M. (2011). "Optimal conditions for multifocal VEP recording for normal Japanese population established by receiver operating characteristic analysis." Documenta Ophthalmologica 122: 29-37.
Janz, C., Heinrich, S.P., Kornmayer, J., Bach, M. and Hennig, J. (2001). "Coupling of neural activity and BOLD fMRI response: new insights by combination of fMRI and VEP experiments in transition from single events to continuous stimulation." Magnetic Resonance in Medicine 46: 482-486.
Jasper, H. (1958). "The ten twenty system of the international federation." Electroencephalography and Clinical Neurophysiology 10: 371-375.
Jaworski, A., Gentle, A., Zele, A.J., Vingrys, A.J. and McBrien, N.A. (2006). "Altered visual sensitivity in axial high myopia: a local postreceptoral phenomenon?" Investigative Ophthalmology and Visual Science 47: 3695-3702.
Johnson, C.A. (1994). "Selective versus nonselective losses in glaucoma." Journal of Glaucoma 3: S32-44.
Judit, A., Sandor, P.S. and Schoenen, J. (2000). "Habituation of visual and intensity dependence of auditory evoked cortical potentials tends to normalize just before and during the migraine attack." Cephalalgia 20: 714-719.
Jueptner, M. and Weiller, C. (1995). "Review: does measurement of regional cerebral blood flow reflect synaptic activity? Implications for PET and fMRI." Neuroimage 2: 148-156.
Kaplan, E. and Shapley, R.M. (1982). "X and Y cells in the lateral geniculate nucleus of macaque monkeys." The Journal of Physiology 330: 125-143.
Kaplan, E., Purpura, K. and Shapley, R.M. (1987). "Contrast affects the transmission of visual information through the mammalian lateral geniculate nucleus." The Journal of Physiology 391: 267-288.
Karanovic, O., Thabet, M., Wilson, H.R. and Wilkinson, F. (2011). "Detection and discrimination of flicker contrast in migraine." Cephalalgia 31: 723-736.
Karas, R. (2012). Perceptual centre-surround processing in older adults. PhD Thesis, The University of Melbourne.
Katz, B. (1986). "Migrainous central retinal artery occlusion." Journal of Clinical Neuroophthalmology 6: 69- 75.
Kaufman, D. and Celesia, G.G. (1985). "Simultaneous recording of pattern electroretinogram and visual evoked responses in neuro-ophthalmologic disorders." Neurology 35: 644-651.
Kelman, L. (2004a). "The aura: a tertiary care study of 952 migraine patients." Cephalalgia 24: 728-734.
Kelman, L. (2004b). "The premonitory symptoms (prodrome): a tertiary care study of 893 migraineurs." Headache 44: 865-872.
Kennard, C., Gawel, M., Rudolph, N. and Rose, F.C. (1978). "Visual evoked potentials in migraine subjects." Research and Clinical Studies in Headache 6: 73-80.
Khalil, N.M. and Legg, N.J. (1989). Pathophysiology of migraine: a study using VEP and contrast sensitivity. In Clifford Rose, R (ed). New Advances in Headache Research. London, United Kingdom, Smith-Gordon & Co.: pp 57-61.
Khalil, N.M. (1991). Investigations of visual function in migraine by visual evoked potentials and visual psychophysical tests, PhD Thesis, University of London, UK.
229 Khalil, N.M., Legg, N.J. and Anderson, D.J. (2000). "Long term decline of P100 amplitude in migraine with aura." Journal of Neurology, Neurosurgery, and Psychiatry 69: 507-511.
Khalil, N.M., Nicotra, A. and Wilkins, A.J. (2011). "Asymmetry of visual function in migraine with aura: correlation with lateralisation of headache and aura." Cephalalgia 31: 213-221.
Killer, H.E., Forrer, A. and Flammer, J. (2003). "Retinal vasospasm during an attack of migraine." Retina 23: 253-254.
Kilpelainen, M., Donner, K. and Laurinen, P. (2007). "Time course of suppression by surround gratings: highly contrast-dependent, but consistently fast." Vision Research 47: 3298-3306.
King-Smith, P.E. and Kulikowski, J.J. (1975). "Pattern and flicker detection analysed by subthreshold summation." The Journal of Physiology 249: 519-548.
King-Smith, P.E., Grigsby, S.S., Vingrys, A.J., Benes, S.C. and Supowit, A. (1994). "Efficient and unbiased modifications of the QUEST threshold method: theory, simulations, experimental evaluation and practical implementation." Vision Research 34: 885-912.
Klistorner, A.I., Graham, S.L., Grigg, J.R. and Billson, F.A. (1998). "Multifocal topographic visual evoked potential: improving objective detection of local visual field defects." Investigative Ophthalmology and Visual Science 39: 937-950.
Klistorner, A.I., Graham, S.L. and Martins, A. (2000). "Multifocal pattern electroretinogram does not demonstrate localised field defects in glaucoma." Documenta Ophthalmologica 100: 155-165.
Knott, J.R. and Irwin, D.A. (1973). "Anxiety, stress, and the contingent negative variation." Archives of General Psychiatry 29: 538-541.
Kobatake, E. and Tanaka, K. (1994). "Neuronal selectivities to complex object features in the ventral visual pathway of the macaque cerebral cortex." Journal of Neurophysiology 71: 856-867.
Koerner, F. and Teuber, H.L. (1973). "Visual field defects after missile injuries to the geniculo-striate pathway in man." Experimental Brain Research 18: 88-113.
Kohn, A. (2007). "Visual adaptation: physiology, mechanisms, and functional benefits." Journal of Neurophysiology 97: 3155-3164.
Korth, M. (1983). "Pattern-evoked responses and luminance-evoked responses in the human electroretinogram." Journal of Physiology 337: 451-469.
Kowacs, P.A., Piovesan, E.J., Werneck, L.C., Fameli, H. and Pereira da Silva, H. (2004). "Headache related to a specific screen flickering frequency band." Cephalalgia 24: 408-410.
Krapin, D. (1964). "Occlusion of the central retinal artery in migraine." New England Journal of Medicine 270: 359-360.
Kruuse, C., Thomsen, L.L., Birk, S. and Olesen, J. (2003). "Migraine can be induced by sildenafil without changes in middle cerebral artery diameter." Brain 126: 241-247.
Kunkler, P.E. and Kraig, R.P. (2003). "Hippocampal spreading depression bilaterally activates the caudal trigeminal nucleus in rodents." Hippocampus 13: 835-844.
Lahat, E., Nadir, E., Barr, J., Eshel, G., Aladjem, M. and Bistritze, T. (1997). "Visual evoked potentials: a diagnostic test for migraine headache in children." Developmental Medicine and Child Neurology 39: 85-87.
230 Lahat, E., Barr, J., Barzilai, A., Cohen, H. and Berkovitch, M. (1999). "Visual evoked potentials in the diagnosis of headache before 5 years of age." European Journal of Pediatrics 158: 892-895.
Lashley, K. (1941). "Patterns of cerebral integration indicated by the scotomas of migraine." Archives of Neurology and Psychiatry 46: 331-339.
Lauritzen, M., Jorgensen, M.B., Diemer, N.H., Gjedde, A. and Hansen, A.J. (1982). "Persistent oligemia of rat cerebral cortex in the wake of spreading depression." Annals of Neurology 12: 469-474.
Lauritzen, M., Skyhoj Olsen, T., Lassen, N.A. and Paulson, O.B. (1983). "Changes in regional cerebral blood flow during the course of classic migraine attacks." Annals of Neurology 13: 633-641.
Lauritzen, M. (1994). "Pathophysiology of the migraine aura. The spreading depression theory." Brain 117: 199-210.
Leão, A. (1944). "Spreading depression of activity in the cerebral cortex." Journal of Neurophysiology 7: 359-390.
Ledgeway, T., Zhan, C., Johnson, A.P., Song, Y. and Baker, C.L., Jr. (2005). "The direction-selective contrast response of area 18 neurons is different for first- and second-order motion." Visual Neuroscience 22: 87-99.
Lehtonen, J.B. (1974). "Visual evoked cortical potentials for single flashes and flickering light in migraine." Headache 14: 1-12.
Lennerstrand, G. (1982). "Delayed visual evoked cortical potentials in retinal disease." Acta Ophthalmologica 60: 497-504.
Leonardi, M. and Raggi, A. (2013). "Burden of migraine: international perspectives." Neurological Sciences 34: S117-118.
Leonova, A., Pokorny, J. and Smith, V.C. (2003). "Spatial frequency processing in inferred PC- and MC-pathways." Vision Research 43: 2133-2139.
Levitt, J.B. and Lund, J.S. (1997). "Contrast dependence of contextual effects in primate visual cortex." Nature 387: 73-76.
Lewis, R.A., Vijayan, N., Watson, C., Keltner, J. and Johnson, C.A. (1989). "Visual field loss in migraine." Ophthalmology 96: 321-326.
Li, C.Y. and Creutzfeldt, O. (1984). "The representation of contrast and other stimulus parameters by single neurons in area 17 of the cat." European Journal of Physiology 401: 304-314.
Lipton, R., Stewart, W., Diamond, S., Diamond, M. and Reed, M. (2001a). "Prevalence and burden of migraine in the United States: Data from the American Migraine Study II." Headache 41: 646-657.
Lipton, R.B., Stewart, W.F., Sawyer, J. and Edmeads, J.G. (2001b). "Clinical utility of an instrument assessing migraine disability: the Migraine Disability Assessment (MIDAS) questionnaire." Headache 41: 854-861.
Livingstone, M. and Hubel, D. (1988). "Segregation of form, color, movement, and depth: anatomy, physiology, and perception." Science 240: 740-749.
Livingstone, M.S. and Hubel, D.H. (1987). "Connections between layer 4B of area 17 and the thick cytochrome oxidase stripes of area 18 in the squirrel monkey." The Journal of Neuroscience 7: 3371-3377.
231 Logi, F., Bonfiglio, L., Orlandi, G., Bonanni, E., Iudice, A. and Sartucci, F. (2001). "Asymmetric scalp distribution of pattern visual evoked potentials during interictal phases in migraine." Acta Neurologica Scandinavica 104: 301-307.
Lorenz, R. and Heider, W. (1990). "Retinal origin of VECP delays as revealed by simultaneously recorded ERG to patterned stimuli." Documenta Ophthalmologica 75: 49-57.
Lovasik, J. and Konietzny, E. (1989). "The effect of retinal image quality on steady state pattern ERGs and VEPs." Clinical Vision Sciences 4: 41-50.
Luo, X. and Frishman, L.J. (2011). "Retinal pathway origins of the pattern electroretinogram (PERG)." Investigative Ophthalmology and Visual Science 52: 8571-8584.
Luu, S.T., Pesudovs, K., Lee, A.W. and Chen, C.S. (2010). "Homonymous hemianopia captured on automated perimetry during a migraine episode." Internal Medicine Journal 40: 310-311.
Lyons, R. (1997). Understanding digital signal processing. Reading, Massachusetts, Addison Wesley Longman, Inc.
Macknik, S.L. and Livingstone, M.S. (1998). "Neuronal correlates of visibility and invisibility in the primate visual system." Nature Neuroscience 1: 144-149.
MacLean, C., Appenzeller, O., Cordaro, J.T. and Rhodes, J. (1975). "Flash evoked potentials in migraine." Headache 14: 193-198.
Maffei, L., Fiorentini, A. and Bisti, S. (1973). "Neural correlate of perceptual adaptation to gratings." Science 182: 1036-1038.
Maffei, L. and Fiorentini, A. (1981). "Electroretinographic responses to alternating gratings before and after section of the optic nerve." Science 211: 953-954.
Magis, D., Ambrosini, A., Bendtsen, L., Ertas, M., Kaube, H. and Schoenen, J. (2007a). "Evaluation and proposal for optimalization of neurophysiological tests in migraine: Part 1 - Electrophysiological tests." Cephalalgia 27: 1323-1338.
Magis, D., Allena, M., Coppola, G., Di Clemente, L., Gerard, P. and Schoenen, J. (2007b). "Search for correlations between genotypes and electrophysiological patterns in migraine: the MTHFR C677T polymorphism and visual evoked potentials." Cephalalgia 27: 1142-1149.
Maher, B.H. and Griffiths, L.R. (2011). "Identification of molecular genetic factors that influence migraine." Molecular Genetics and Genomics 285: 433-446.
Manzoni, G.C. and Torelli, P. (2008). "Migraine with and without aura: a single entity?" Neurological Sciences 29: S40-43.
Marcus, D.A. and Soso, M.J. (1989). "Migraine and stripe-induced visual discomfort." Archives of Neurology 46: 1129-1132.
Mariani, E., Moschini, V., Pastorino, G.C., Rizzi, F., Severgnini, A. and Tiengo, M. (1990). "Pattern reversal visual evoked potentials (VEP-PR) in migraine subjects with visual aura." Headache 30: 435- 438.
Marrelli, A., Tozzi, E., Porto, C., Cimini, N., Aloisi, P. and Valenti, M. (2001). "Spectral analysis of visual potentials evoked by pattern-reversal checkerboard in juvenile patients with headache." Headache 41: 792-797.
232 Martinez, A., Proupim, N. and Sanchez, M. (2008). "Retinal nerve fibre layer thickness measurements using optical coherence tomography in migraine patients." British Journal of Ophthalmology 92: 1069- 1075.
Martinez, A., Proupim, N. and Sanchez, M. (2009). "Scanning laser polarimetry with variable corneal compensation in migraine patients." Acta Ophthalmologica 87: 746-753.
Matharu, M.S., Bartsch, T., Ward, N., Frackowiak, R.S., Weiner, R. and Goadsby, P.J. (2004). "Central neuromodulation in chronic migraine patients with suboccipital stimulators: a PET study." Brain 127: 220-230.
Matsui, Y., Saito, I., Okinami, S. and Oono, S. (1994). "Influence of simultaneous pattern-reversal electroretinogram recording on visual evoked potentials." Documenta Ophthalmologica 86: 285-294.
Maunsell, J.H., Nealey, T.A. and DePriest, D.D. (1990). "Magnocellular and parvocellular contributions to responses in the middle temporal visual area (MT) of the macaque monkey." The Journal of Neuroscience 10: 3323-3334.
Mazzucchi, A., Sinforiani, E., Zinelli, P., Agostinis, C., Granella, F., Miari, A., Manzoni, G.C. and Parma, M. (1988). "Interhemispheric attentional functioning in classic migraine subjects during paroxysmal and interparoxysmal phases." Headache 28: 488-493.
McColl, S. and Wilkinson, F. (2001). "Migraineurs show visual field defects on motion coherence perimetry during the interictal period. Abstract of the 10th Congress of the International Headache Society, New York, NY, June 2001." Cephalalgia 21: 391.
McColl, S.L. and Wilkinson, F. (2000). "Visual contrast gain control in migraine: measures of visual cortical excitability and inhibition." Cephalalgia 20: 74-84.
McCulloch, D.L., Van Boemel, G.B. and Borchert, M.S. (1997). "Comparisons of contact lens, foil, fiber and skin electrodes for patterns electroretinograms." Documenta Ophthalmologica 94: 327-340.
McKendrick, A.M., Badcock, D.R., Heywood, J. and Vingrys, A.J. (1998). "Effects of migraine on visual function." Australian and New Zealand Journal of Ophthalmology 26: S111-113.
McKendrick, A.M. (1999). Vision after migraine. PhD Thesis, The University of Melbourne
McKendrick, A.M., Vingrys, A.J., Badcock, D.R. and Heywood, J.T. (2000). "Visual field losses in subjects with migraine headaches." Investigative Ophthalmology and Visual Science 41: 1239-1247.
McKendrick, A.M., Vingrys, A.J., Badcock, D.R. and Heywood, J.T. (2001). "Visual dysfunction between migraine events." Investigative Ophthalmology and Visual Science 42: 626-633.
McKendrick, A.M., Cioffi, G.A. and Johnson, C.A. (2002). "Short-wavelength sensitivity deficits in patients with migraine." Archives of Ophthalmology 120: 154-161.
McKendrick, A.M. and Badcock, D.R. (2003). "Contrast-processing dysfunction in both magnocellular and parvocellular pathways in migraineurs with or without aura." Investigative Ophthalmology and Visual Science 44: 442-448.
McKendrick, A.M. and Badcock, D.R. (2004a). "An analysis of the factors associated with visual field deficits measured with flickering stimuli in-between migraine." Cephalalgia 24: 389-397.
McKendrick, A.M. and Badcock, D.R. (2004b). "Decreased visual field sensitivity measured 1 day, then 1 week, after migraine." Investigative Ophthalmology and Visual Science 45: 1061-1070.
233 McKendrick, A.M. and Badcock, D.R. (2004c). "Motion processing deficits in migraine." Cephalalgia 24: 363-372.
McKendrick, A.M., Badcock, D.R., Badcock, J.C. and Gurgone, M. (2006a). "Motion perception in migraineurs: abnormalities are not related to attention." Cephalalgia 26: 1131-1136.
McKendrick, A.M., Badcock, D.R. and Gurgone, M. (2006b). "Vernier acuity is normal in migraine, whereas global form and global motion perception are not." Investigative Ophthalmology and Visual Science 47: 3213-3219.
McKendrick, A.M. and Sampson, G.P. (2009). "Low spatial frequency contrast sensitivity deficits in migraine are not visual pathway selective." Cephalalgia 29: 539-549.
McKendrick, A.M., Battista, J., Snyder, J.S. and Carter, O.L. (2011). "Visual and auditory perceptual rivalry in migraine." Cephalalgia 31: 1158-1169.
Medmont M700 Automated Perimeter User Manual (2007). Vermont, Medmont International Pty Ltd: pp 1-115.
Meigen, T. and Bach, M. (1999). "On the statistical significance of electrophysiological steady-state responses." Documenta Ophthalmologica 98: 207-232.
Merigan, W.H. and Eskin, T.A. (1986). "Spatio-temporal vision of macaques with severe loss of P beta retinal ganglion cells." Vision Research 26: 1751-1761.
Merigan, W.H. and Maunsell, J.H. (1990). "Macaque vision after magnocellular lateral geniculate lesions." Visual Neuroscience 5: 347-352.
Merigan, W.H., Katz, L.M. and Maunsell, J.H. (1991). "The effects of parvocellular lateral geniculate lesions on the acuity and contrast sensitivity of macaque monkeys." The Journal of Neuroscience 11: 994- 1001.
Merigan, W.H., Nealey, T.A. and Maunsell, J.H. (1993). "Visual effects of lesions of cortical area V2 in macaques." The Journal of Neuroscience 13: 3180-3191.
Metha, A., Vingrys, A. and Badcock, D. (1993). "Calibration of a color monitor for visual psychophysics." Behavior Research Methods Instruments Computers 25: 371-383.
Mezer, E., Bahir, Y., Leibu, R. and Perlman, I. (2004). "Effect of defocusing and of distracted attention upon recordings of the visual evoked potential." Documenta Ophthalmologica 109: 229-238.
Michelson, A. (1927). Studies in Optics. Chicago, University of Chicago Press.
Moehnke, T.D., Sowka, J., Shallo-Hoffmann, J., Hardigan, P. and Woods, A.D. (2008). "Topographical analysis of the optic nerve in migraine patients." Optometry and Vision Science 85: 566- 573.
Monteith, T.S. and Goadsby, P.J. (2011). "Acute migraine therapy: new drugs and new approaches." Current Treatment Options in Neurology 13: 1-14.
Moreira Filho, P. and Dantas, A. (1995). "[Electroretinography by reversal pattern in patients with migraine without aura]." Arquivos de Neuro-Psiquiatria 53: 598-603.
Morrone, M.C. and Burr, D.C. (1986). "Evidence for the existence and development of visual inhibition in humans." Nature 321: 235-237.
234 Morrone, M.C., Burr, D.C. and Speed, H.D. (1987). "Cross-orientation inhibition in cat is GABA mediated." Experimental Brain Research 67: 635-644.
Moskowitz, M.A. and Macfarlane, R. (1993). "Neurovascular and molecular mechanisms in migraine headaches." Cerebrovascular and Brain Metabolism Reviews 5: 159-177.
Mulder, E.J., Linssen, W.H., Passchier, J., Orlebeke, J.F. and de Geus, E.J. (1999). "Interictal and postictal cognitive changes in migraine." Cephalalgia 19: 557-565.
Mulleners, W.M., Aurora, S.K., Chronicle, E.P., Stewart, R., Gopal, S. and Koehler, P.J. (2001a). "Self- reported photophobic symptoms in migraineurs and controls are reliable and predict diagnostic category accurately." Headache 41: 31-39.
Mulleners, W.M., Chronicle, E.P., Palmer, J.E., Koehler, P.J. and Vredeveld, J.W. (2001b). "Suppression of perception in migraine: evidence for reduced inhibition in the visual cortex." Neurology 56: 178-183.
Newsome, W.T. and Pare, E.B. (1988). "A selective impairment of motion perception following lesions of the middle temporal visual area (MT)." The Journal of Neuroscience 8: 2201-2211.
Newsome, W.T., Britten, K.H. and Movshon, J.A. (1989). "Neuronal correlates of a perceptual decision." Nature 341: 52-54.
Nizankowska, M.H., Turno-Krecicka, A., Misiuk-Hojlo, M., Ejma, M., Chelstowska, J., Szczesna- Borzemska, D. and Sasiadek, M. (1997). "[Coexistence of migraine and glaucoma-like visual field defects]." Klinika Oczna 99: 121-126.
Noseda, R. and Burstein, R. (2013). "Migraine pathophysiology: anatomy of the trigeminovascular pathway and associated neurological symptoms, CSD, sensitization and modulation of pain." Pain 154: S44-53.
Nyrke, T. and Lang, A.H. (1982). "Spectral analysis of visual potentials evoked by sine wave modulated light in migraine." Electroencephalography and Clinical Neurophysiology 53: 436-442.
Nyrke, T., Kangasniemi, P. and Lang, A.H. (1989). "Difference of steady-state visual evoked potentials in classic and common migraine." Electroencephalography and Clinical Neurophysiology 73: 285- 294.
O'Bryant, S.E., Marcus, D.A., Rains, J.C. and Penzien, D.B. (2005). "Neuropsychology of migraine: present status and future directions." Expert Review of Neurotherapeutics 5: 363-370.
O'Keeffe, S.T., Tsapatsaris, N.P. and Beetham, W.P., Jr. (1993). "Association between Raynaud's phenomenon and migraine in a random population of hospital employees." Journal of Rheumatology 20: 1187-1188.
Obrig, H., Israel, H., Kohl-Bareis, M., Uludag, K., Wenzel, R., Muller, B., Arnold, G. and Villringer, A. (2002). "Habituation of the visually evoked potential and its vascular response: implications for neurovascular coupling in the healthy adult." Neuroimage 17: 1-18.
Odom, J.V., Maida, T.M. and Dawson, W.W. (1982). "Pattern evoked retinal response (PERR) in human: effects of spatial frequency, temporal frequency, luminance and defocus." Current Eye Research 2: 99-108.
Odom, J.V., Holder, G.E., Feghali, J.G. and Cavender, S. (1992). "Pattern electroretinogram intrasession reliability: a two center comparison." Clinical Vision Sciences 7: 263-281.
235 Odom, J.V., Bach, M., Brigell, M., Holder, G.E., McCulloch, D.L., Tormene, A.P. and Vaegan (2010). "ISCEV standard for clinical visual evoked potentials." Documenta Ophthalmologica 120: 111-119.
Oelkers, R., Grosser, K., Lang, E., Geisslinger, G., Kobal, G., Brune, K. and Lotsch, J. (1999). "Visual evoked potentials in migraine patients: alterations depend on pattern spatial frequency." Brain 122: 1147-1155.
Oelkers-Ax, R., Parzer, P., Resch, F. and Weisbrod, M. (2005). "Maturation of early visual processing investigated by a pattern-reversal habituation paradigm is altered in migraine." Cephalalgia 25: 280-289.
Ohzawa, I., Sclar, G. and Freeman, R.D. (1985). "Contrast gain control in the cat's visual system." Journal of Neurophysiology 54: 651-667.
Olesen, J., Larsen, B. and Lauritzen, M. (1981). "Focal hyperaemia followed by spreading oligemia and impaired activation of rCBF in classic migraine." Annals of Neurology 9: 344-352.
Omland, P.M., Nilsen, K.B. and Sand, T. (2011). "Habituation measured by pattern reversal visual evoked potentials depends more on check size than reversal rate." Clinical Neurophysiology 122: 1846- 1853.
Omland, P.M., Nilsen, K.B., Uglem, M., Gravdahl, G., Linde, M., Hagen, K. and Sand, T. (2013). "Visual evoked potentials in interictal migraine: no confirmation of abnormal habituation." Headache 53: 1071-1086.
Orgul, S. and Flammer, J. (1994). "Headache in normal-tension glaucoma patients." Journal of Glaucoma 3: 292-295.
Ozeki, H., Sadakane, O., Akasaki, T., Naito, T., Shimegi, S. and Sato, H. (2004). "Relationship between excitation and inhibition underlying size tuning and contextual response modulation in the cat primary visual cortex." The Journal of Neuroscience 24: 1428-1438.
Ozkul, Y. and Bozlar, S. (2002). "Effects of fluoxetine on habituation of pattern reversal visually evoked potentials in migraine prophylaxis." Headache 42: 582-587.
Pack, C.C., Berezovskii, V.K. and Born, R.T. (2001). "Dynamic properties of neurons in cortical area MT in alert and anaesthetized macaque monkeys." Nature 414: 905-908.
Pack, C.C., Hunter, J.N. and Born, R.T. (2005). "Contrast dependence of suppressive influences in cortical area MT of alert macaque." Journal of Neurophysiology 93: 1809-1815.
Palermo, A., Giglia, G., Vigneri, S., Cosentino, G., Fierro, B. and Brighina, F. (2011). "Does habituation depend on cortical inhibition? Results of a rTMS study in healthy subjects." Experimental Brain Research 212: 101-107.
Palmer, J.E., Chronicle, E.P., Rolan, P. and Mulleners, W.M. (2000). "Cortical hyperexcitability is cortical under-inhibition: evidence from a novel functional test of migraine patients." Cephalalgia 20: 525-532.
Panconesi, A. (2008). "Serotonin and migraine: a reconsideration of the central theory." The Journal of Headache and Pain 9: 267-276.
Paradis, A.L., Cornilleau-Peres, V., Droulez, J., Van De Moortele, P.F., Lobel, E., Berthoz, A., Le Bihan, D. and Poline, J.B. (2000). "Visual perception of motion and 3-D structure from motion: an fMRI study." Cerebral Cortex 10: 772-783.
236 Parisi, V., Miglior, S., Manni, G., Centofanti, M. and Bucci, M.G. (2006). "Clinical ability of pattern electroretinograms and visual evoked potentials in detecting visual dysfunction in ocular hypertension and glaucoma." Ophthalmology 113: 216-228.
Pastor, M.A., Artieda, J., Arbizu, J., Valencia, M. and Masdeu, J.C. (2003). "Human cerebral activation during steady-state visual-evoked responses." Journal of Neuroscience 23: 11621-11627.
Peirce, J.W. (2007). "The potential importance of saturating and supersaturating contrast response functions in visual cortex." Journal of Vision 7: 1-10.
Percival, K.A., Martin, P.R. and Grunert, U. (2013). "Organisation of koniocellular-projecting ganglion cells and diffuse bipolar cells in the primate fovea." European Journal of Neuroscience 37: 1072- 1089.
Perry, V.H., Oehler, R. and Cowey, A. (1984). "Retinal ganglion cells that project to the dorsal lateral geniculate nucleus in the macaque monkey." Neuroscience 12: 1101-1123.
Petrov, Y., Carandini, M. and McKee, S. (2005). "Two distinct mechanisms of suppression in human vision." The Journal of Neuroscience 25: 8704-8707.
Petrov, Y. and McKee, S.P. (2009). "The time course of contrast masking reveals two distinct mechanisms of human surround suppression." Journal of Vision 9: 1-11.
Pfeiffer, N. and Bach, M. (1992). "The pattern-electroretinogram in glaucoma and ocular hypertension. A cross-sectional and longitudinal study." German Journal of Ophthalmology 1: 35-40.
Phelps, C.D. and Corbett, J.J. (1985). "Migraine and low-tension glaucoma. A case-control study." Investigative Ophthalmology and Visual Science 26: 1105-1108.
Pietrobon, D. (2005). "Migraine: new molecular mechanisms." The Neuroscientist 11: 373-386.
Pietrobon, D. and Moskowitz, M.A. (2013). "Pathophysiology of migraine." Annual Review of Physiology 75: 365-391.
Plange, N., Remky, A. and Arend, O. (2003). "Colour Doppler imaging and fluorescein filling defects of the optic disc in normal tension glaucoma." British Journal of Ophthalmology 87: 731-736.
Pokorny, J. and Smith, V.C. (1997). "Psychophysical signatures associated with magnocellular and parvocellular pathway contrast gain." Journal of the Optical Society of America. A. Optics, Image Science, and Vision 14: 2477-2486.
Polich, J., Ehlers, C.L. and Dalessio, D.J. (1986). "Pattern-shift visual evoked responses and EEG in migraine." Headache 26: 451-456.
Porciatti, V., Bonanni, P., Fiorentini, A. and Guerrini, R. (2000). "Lack of cortical contrast gain control in human photosensitive epilepsy." Nature Neuroscience 3: 259-263.
Porciatti, V. and Ventura, L.M. (2004). "Normative data for a user-friendly paradigm for pattern electroretinogram recording." Ophthalmology 111: 161-168.
Porciatti, V., Sorokac, N. and Buchser, W. (2005). "Habituation of retinal ganglion cell activity in response to steady state pattern visual stimuli in normal subjects." Investigative Ophthalmology and Visual Science 46: 1296-1302.
Porciatti, V. and Ventura, L.M. (2009a). "Adaptive changes of inner retina function in response to sustained pattern stimulation." Vision Research 49: 505-513.
237 Porciatti, V. and Ventura, L.M. (2009b). "Physiologic significance of steady-state pattern electroretinogram losses in glaucoma: clues from simulation of abnormalities in normal subjects." Journal of Glaucoma 18: 535-542.
Pradalier, A., Hamard, P., Sellem, E. and Bringer, L. (1998). "Migraine and glaucoma: an epidemiologic survey of French ophthalmologists." Cephalalgia 18: 74-76.
Price, M.J., Drance, S.M., Price, M., Schulzer, M., Douglas, G.R. and Tansley, B. (1988). "The pattern electroretinogram and visual-evoked potential in glaucoma." Graefes Archives of Clinical and Experimental Ophthalmology 226: 542-547.
Puppe, A. and Limmroth, V. (2007). "GABAergic drugs for the treatment of migraine." Central Nervous System and Neurological Disorders. Drug Targets 6: 247-250.
Purdy, R.A. (2008). "Migraine with and without aura share the same pathogenic mechanisms." Neurological Sciences 29: S44-46.
Rahmann, A., Wienecke, T., Hansen, J.M., Fahrenkrug, J., Olesen, J. and Ashina, M. (2008). "Vasoactive intestinal peptide causes marked cephalic vasodilation, but does not induce migraine." Cephalalgia 28: 226-236.
Rasmussen, B.K. and Olesen, J. (1992). "Migraine with aura and migraine without aura: an epidemiological study." Cephalalgia 12: 221-228.
Razeghinejad, M.R., Masoumpour, M. and Bagheri, M.H. (2009). "Migrainous prolonged and reversible bilateral inferior altitudinal visual field defect." Headache 49: 773-776.
Regan, D. (1977). "Steady-state evoked potentials." Journal of the Optical Society of America 67: 1475-1489.
Regan, D. (1982). "Comparison of transient and steady-state methods." Annals of the New York Academy of Sciences 388: 45-71.
Rich, W.M. (1948). "Permanent homonymous quadrantanopia after migraine." British Medical Journal 1: 592-594.
Riva, C.E., Falsini, B. and Logean, E. (2001). "Flicker-evoked responses of human optic nerve head blood flow: luminance versus chromatic modulation." Investigative Ophthalmology and Visual Science 42: 756-762.
Riva, C.E., Logean, E. and Falsini, B. (2005). "Visually evoked hemodynamical response and assessment of neurovascular coupling in the optic nerve and retina." Progress in Retinal and Eye Research 24: 183-215.
Rodieck, R.W. and Watanabe, M. (1993). "Survey of the morphology of macaque retinal ganglion cells that project to the pretectum, superior colliculus, and parvicellular laminae of the lateral geniculate nucleus." The Journal of Comparative Neurology 338: 289-303.
Rose, K.M., Carson, A.P., Sanford, C.P., Stang, P.E., Brown, C.A., Folsom, A.R. and Szklo, M. (2004). "Migraine and other headaches: associations with Rose angina and coronary heart disease." Neurology 63: 2233-2239.
Ross, J. and Speed, H.D. (1991). "Contrast adaptation and contrast masking in human vision." Proceedings of the Royal Society of London. Series B: Biological Sciences 246: 61-69.
238 Rossi, L.N., Pastorino, G.C., Bellettini, G., Chiodi, A., Mariani, E. and Cortinovis, I. (1996). "Pattern reversal visual evoked potentials in children with migraine or tension-type headache." Cephalalgia 16: 104-106.
Russell, M.B., Iversen, H.K. and Olesen, J. (1994). "Improved description of the migraine aura by a diagnostic aura diary." Cephalalgia 14: 107-117.
Russell, M.B. and Olesen, J. (1996). "A nosographic analysis of the migraine aura in a general population." Brain 119: 355-361.
Sakai, Y., Dobson, C., Diksic, M., Aube, M. and Hamel, E. (2008). "Sumatriptan normalizes the migraine attack-related increase in brain serotonin synthesis." Neurology 70: 431-439.
Sakaue, H., Katsumi, O., Mehta, M. and Hirose, T. (1990). "Simultaneous pattern reversal ERG and VER recordings. Effect of stimulus field and central scotoma." Investigative Ophthalmology and Visual Science 31: 506-511.
Sample, P.A. and Weinreb, R.N. (1990). "Color perimetry for assessment of primary open-angle glaucoma." Investigative Ophthalmology and Visual Science 31: 1869-1875.
Sanabria, O., Feuer, W.J. and Anderson, D.R. (1991). "Pseudo-loss of fixation in automated perimetry." Ophthalmology 98: 76-78.
Sanchez del Rio, M., Bakker, D., Wu, O., Agosti, R., Mitsikostas, D.D., Ostergaard, L., Wells, W.A., Rosen, B.R., Sorensen, G., Moskowitz, M.A. and Cutrer, F.M. (1999). "Perfusion weighted imaging during migraine: spontaneous visual aura and headache." Cephalalgia 19: 701-707.
Sanchez-Vives, M.V., Nowak, L.G. and McCormick, D.A. (2000a). "Cellular mechanisms of long- lasting adaptation in visual cortical neurons in vitro." The Journal of Neuroscience 20: 4286-4299.
Sanchez-Vives, M.V., Nowak, L.G. and McCormick, D.A. (2000b). "Membrane mechanisms underlying contrast adaptation in cat area 17 in vivo." The Journal of Neuroscience 20: 4267-4285.
Sand, T. and Vingen, J.V. (2000). "Visual, long-latency auditory and brainstem auditory evoked potentials in migraine: relation to pattern size, stimulus intensity, sound and light discomfort thresholds and pre-attack state." Cephalalgia 20: 804-820.
Sand, T., Zhitniy, N., White, L.R. and Stovner, L.J. (2008). "Visual evoked potential latency, amplitude and habituation in migraine: a longitudinal study." Clinical Neurophysiology 119: 1020-1027.
Sand, T., White, L.R., Hagen, K. and Stovner, L.J. (2009). "Visual evoked potential and spatial frequency in migraine: a longitudinal study." Acta Neurologica Scandinavica 120: 33-37.
Sandor, P.S., Dydak, U., Schoenen, J., Kollias, S.S., Hess, K., Boesiger, P. and Agosti, R.M. (2005). "MR-spectroscopic imaging during visual stimulation in subgroups of migraine with aura." Cephalalgia 25: 507-518.
Sappey-Marinier, D., Calabrese, G., Fein, G., Hugg, J.W., Biggins, C. and Weiner, M.W. (1992). "Effect of photic stimulation on human visual cortex lactate and phosphates using 1H and 31P magnetic resonance spectroscopy." Journal of Cerebral Blood Flow and Metabolism 12: 584-592.
Sarchielli, P., Tarducci, R., Presciutti, O., Gobbi, G., Pelliccioli, G.P., Stipa, G., Alberti, A. and Capocchi, G. (2005). "Functional 1H-MRS findings in migraine patients with and without aura assessed interictally." Neuroimage 24: 1025-1031.
239 Schiller, P.H. and Malpeli, J.G. (1978). "Functional specificity of lateral geniculate nucleus laminae of the rhesus monkey." Journal of Neurophysiology 41: 788-797.
Schiller, P.H. (1993). "The effects of V4 and middle temporal (MT) area lesions on visual performance in the rhesus monkey." Visual Neuroscience 10: 717-746.
Schmitz, N., Admiraal-Behloul, F., Arkink, E.B., Kruit, M.C., Schoonman, G.G., Ferrari, M.D. and van Buchem, M.A. (2008). "Attack frequency and disease duration as indicators for brain damage in migraine." Headache 48: 1044-1055.
Schoenen, J., Wang, W., Albert, A. and Delwaide, P. (1995). "Potentiation instead of habituation characterizes visual evoked potentials in migraine patients between attacks." European Journal of Neurology 2: 115-122.
Schoenen, J. (1996). "Deficient habituation of evoked cortical potentials in migraine: a link between brain biology, behavior and trigeminovascular activation?" Biomedicine and Pharmacotherapy 50: 71-78.
Schoonman, G.G., Evers, D.J., Terwindt, G.M., van Dijk, J.G. and Ferrari, M.D. (2006). "The prevalence of premonitory symptoms in migraine: a questionnaire study in 461 patients." Cephalalgia 26: 1209-1213.
Schoonman, G.G., van der Grond, J., Kortmann, C., van der Geest, R.J., Terwindt, G.M. and Ferrari, M.D. (2008). "Migraine headache is not associated with cerebral or meningeal vasodilatation--a 3T magnetic resonance angiography study." Brain 131: 2192-2200.
Schott, G.D. (2007). "Exploring the visual hallucinations of migraine aura: the tacit contribution of illustration." Brain 130: 1690-1703.
Schurks, M., Rist, P.M., Bigal, M.E., Buring, J.E., Lipton, R.B. and Kurth, T. (2009). "Migraine and cardiovascular disease: systematic review and meta-analysis." British Medical Journal 339: b3914.
Schwabe, L., Ichida, J.M., Shushruth, S., Mangapathy, P. and Angelucci, A. (2010). "Contrast- dependence of surround suppression in macaque V1. Experimental testing of a recurrent network model." Neuroimage 52: 777-792.
Schwedt, T.J. and Dodick, D.W. (2009). "Advanced neuroimaging of migraine." The Lancet Neurology 8: 560-568.
Sclar, G., Lennie, P. and DePriest, D.D. (1989). "Contrast adaptation in striate cortex of macaque." Vision Research 29: 747-755.
Sclar, G., Maunsell, J.H. and Lennie, P. (1990). "Coding of image contrast in central visual pathways of the macaque monkey." Vision Research 30: 1-10.
Sener, H.O., Haktanir, I. and Demirci, S. (1997). "Pattern-reversal visual evoked potentials in migraineurs with or without visual aura." Headache 37: 449-451.
Shepheard, S., Edvinsson, L., Cumberbatch, M., Williamson, D., Mason, G., Webb, J., Boyce, S., Hill, R. and Hargreaves, R. (1999). "Possible antimigraine mechanisms of action of the 5HT1F receptor agonist LY334370." Cephalalgia 19: 851-858.
Shepherd, A.J. (2000). "Visual contrast processing in migraine." Cephalalgia 20: 865-880.
Shepherd, A.J. (2001). "Increased visual after-effects following pattern adaptation in migraine: a lack of intracortical excitation?" Brain 124: 2310-2318.
240 Shepherd, A.J., Palmer, J.E. and Davis, G. (2002). "Increased visual after-effects in migraine following pattern adaptation extend to simultaneous tilt illusion." Spatial Vision 16: 33-43.
Shepherd, A.J. (2005). "Colour vision in migraine: selective deficits for S-cone discriminations." Cephalalgia 25: 412-423.
Shepherd, A.J. (2006). "Local and global motion after-effects are both enhanced in migraine, and the underlying mechanisms differ across cortical areas." Brain 129: 1833-1843.
Shepherd, A.J., Wyatt, G. and Tibber, M.S. (2011). "Visual metacontrast masking in migraine." Cephalalgia 31: 346-356.
Shepherd, A.J., Beaumont, H.M. and Hine, T.J. (2012). "Motion processing deficits in migraine are related to contrast sensitivity." Cephalalgia 32: 554-570.
Shibata, K., Osawa, M. and Iwata, M. (1997). "Simultaneous recording of pattern reversal electroretinograms and visual evoked potentials in migraine." Cephalalgia 17: 742-747.
Shibata, K., Osawa, M. and Iwata, M. (1998). "Pattern reversal visual evoked potentials in migraine with aura and migraine aura without headache." Cephalalgia 18: 319-323.
Shibata, K., Yamane, K., Otuka, K. and Iwata, M. (2008). "Abnormal visual processing in migraine with aura: a study of steady-state visual evoked potentials." Journal of the Neurological Sciences 271: 119- 126.
Shibata, K., Yamane, K., Nishimura, Y., Kondo, H. and Otuka, K. (2011). "Spatial frequency differentially affects habituation in migraineurs: a steady-state visual-evoked potential study." Documenta Ophthalmologica 123: 65-73.
Shields, K.G. and Goadsby, P.J. (2006). "Serotonin receptors modulate trigeminovascular responses in ventroposteromedial nucleus of thalamus: a migraine target?" Neurobiology of Disease 23: 491-501.
Sicuteri, F. (1972). "Headache as possible expression of deficiency of brain 5-hydroxytryptamine (central denervation supersensitivity)." Headache 12: 69-72.
Sieving, P.A. and Steinberg, R.H. (1987). "Proximal retinal contribution to the intraretinal 8-Hz pattern ERG of cat." Journal of Neurophysiology 57: 104-120.
Simpson, T. (1991). "The suppression effect of simulated anisometropia." Ophthalmic and Physiological Optics 11: 350-358.
Singh, M., Kim, S. and Kim, T.S. (2003). "Correlation between BOLD-fMRI and EEG signal changes in response to visual stimulus frequency in humans." Magnetic Resonance in Medicine 49: 108-114.
Siniatchkin, M., Averkina, N., Andrasik, F., Stephani, U. and Gerber, W.D. (2006). "Neurophysiological reactivity before a migraine attack." Neuroscience Letters 400: 121-124.
Siniatchkin, M., Reich, A.L., Shepherd, A.J., van Baalen, A., Siebner, H.R. and Stephani, U. (2009). "Peri-ictal changes of cortical excitability in children suffering from migraine without aura." Pain 147: 132-140.
Skyhoj Olsen, T., Friberg, L. and Lassen, N.A. (1987). "Ischemia may be the primary cause of the neurologic deficits in classic migraine." Archives of Neurology 44: 156-161.
241 Solomon, S.G., White, A.J. and Martin, P.R. (2002). "Extraclassical receptive field properties of parvocellular, magnocellular, and koniocellular cells in the primate lateral geniculate nucleus." The Journal of Neuroscience 22: 338-349.
Souza, G., Hamer, R., Costa, T., Araujo, C., da Silva, V., Gomes, B., Ventura, D. and Silveira, L. (2012). "Contrast processing mechanisms in human visual cortex assessed by visual evoked potentials." Journal of Medicine and Medical Sciences 3: 831-837.
Spekreijse, H., van der Twell, L.H. and Zuidema, T. (1973). "Contrast evoked responses in man." Vision Research 13: 1577-1601.
Spreafico, C., Frigerio, R., Santoro, P., Ferrarese, C. and Agostoni, E. (2004). "Visual evoked potentials in migraine." Neurological Sciences 25: S288-290.
Steinmetz, H., Furst, G. and Freund, H.J. (1989). "Cerebral cortical localization: application and validation of the proportional grid system in MR imaging." Journal of Computer Assisted Tomography 13: 10-19.
Stewart, W., Shechter, A. and Rasmussen, B. (1994). "Migraine prevalence. A review of population based studies." Neurology 44: S17-S23.
Stewart, W.F., Lipton, R.B., Celentano, D.D. and Reed, M.L. (1992). "Prevalence of migraine headache in the United States. Relation to age, income, race, and other sociodemographic factors." The Journal of the American Medical Association 267: 64-69.
Stewart, W.F., Lipton, R.B., Dowson, A.J. and Sawyer, J. (2001). "Development and testing of the Migraine Disability Assessment (MIDAS) Questionnaire to assess headache-related disability." Neurology 56: S20-28.
Stovner, L., Hagen, K., Jensen, R., Katsarava, Z., Lipton, R., Scher, A., Steiner, T. and Zwart, J.A. (2007). "The global burden of headache: a documentation of headache prevalence and disability worldwide." Cephalalgia 27: 193-210.
Strasburger, H., Murray, I. and Remky, A. (1993). "Sustained and transient mechanisms in the steady- state visual evoked potential: onset presentation compared to pattern reversal." Clinical Vision Sciences 8: 211-234.
Sullivan-Mee, M. and Bowman, B. (1997). "Migraine-related visual-field loss with prolonged recovery." Journal of the American Optometric Association 68: 377-388.
Swanson, W.H., Sun, H., Lee, B.B. and Cao, D. (2011). "Responses of primate retinal ganglion cells to perimetric stimuli." Investigative Ophthalmology and Visual Science 52: 764-771.
Szmajda, B.A., Grunert, U. and Martin, P.R. (2008). "Retinal ganglion cell inputs to the koniocellular pathway." Journal of Comparative Neurology 510: 251-268.
Tadin, D., Lappin, J.S., Gilroy, L.A. and Blake, R. (2003). "Perceptual consequences of centre- surround antagonism in visual motion processing." Nature 424: 312-315.
Tadin, D., Kim, J., Doop, M.L., Gibson, C., Lappin, J.S., Blake, R. and Park, S. (2006). "Weakened center-surround interactions in visual motion processing in schizophrenia." The Journal of Neuroscience 26: 11403-11412.
Tagliati, M., Sabbadini, M., Bernardi, G. and Silvestrini, M. (1995). "Multichannel visual evoked potentials in migraine." Electroencephalography and Clinical Neurophysiology 96: 1-5.
242 Tan, F.U., Akarsu, C. and Gullu, R. (2005). "Retinal nerve fiber layer thickness is unaffected in migraine patients." Acta Neurologica Scandinavica 112: 19-23.
Thabet, M., Wilkinson, F., Wilson, H.R. and Karanovic, O. (2013). "The locus of flicker adaptation in the migraine visual system: a dichoptic study." Cephalalgia 33: 5-19.
Tibber, M.S., Guedes, A. and Shepherd, A.J. (2006). "Orientation discrimination and contrast detection thresholds in migraine for cardinal and oblique angles." Investigative Ophthalmology and Visual Science 47: 5599-5604.
Tibber, M.S. and Shepherd, A.J. (2006). "Transient tritanopia in migraine: evidence for a large-field retinal abnormality in blue-yellow opponent pathways." Investigative Ophthalmology and Visual Science 47: 5125-5131.
Tietjen, G.E. (2009). "Migraine as a systemic vasculopathy." Cephalalgia 29: 987-996.
Tobimatsu, S., Celesia, G.G., Cone, S. and Gujrati, M. (1989). "Electroretinograms to checkerboard pattern reversal in cats: Physiological characteristics and effect of retrograde degeneration of ganglion cells." Electroencephalography and Clinical Neurophysiology 73: 341-352.
Tootell, R.B., Hadjikhani, N.K., Vanduffel, W., Liu, A.K., Mendola, J.D., Sereno, M.I. and Dale, A.M. (1998). "Functional analysis of primary visual cortex (V1) in humans." Proceedings of the National Academy of Sciences of the United States of America 95: 811-817.
Trick, G.L. (1985). "Retinal potentials in patients with primary open-angle glaucoma: physiological evidence for temporal frequency tuning deficits." Investigative Ophthalmology and Visual Science 26: 1750- 1758.
Truchard, A.M., Ohzawa, I. and Freeman, R.D. (2000). "Contrast gain control in the visual cortex: monocular versus binocular mechanisms." The Journal of Neuroscience 20: 3017-3032.
Tsai, J.J., Norcia, A.M., Ales, J.M. and Wade, A.R. (2011). "Contrast gain control abnormalities in idiopathic generalized epilepsy." Annals of Neurology 70: 574-582.
Tsai, J.J., Wade, A.R. and Norcia, A.M. (2012). "Dynamics of normalization underlying masking in human visual cortex." The Journal of Neuroscience 32: 2783-2789.
Tsai, T. (2012). Functional correlates between the rat electroretinogram and visual evoked potential. PhD Thesis, University of Melbourne.
Tsaousis, K.T., Plainis, S., Parry, N.R., Pallikaris, I.G., Tsilimbaris, M.K. and Detorakis, E.T. (2013). "Visual electrodiagnosis in glaucoma screening: a clinical study." Journal of Glaucoma 22: 427-431.
Tsounis, S., Milonas, J. and Gilliam, F. (1993). "Hemi-field pattern reversal visual evoked potentials in migraine." Cephalalgia 13: 267-271.
Tsumoto, T., Eckart, W. and Creutzfeldt, O.D. (1979). "Modification of orientation sensitivity of cat visual cortex neurons by removal of GABA-mediated inhibition." Experimental Brain Research 34: 351- 363.
Turpin, A., McKendrick, A.M., Johnson, C.A. and Vingrys, A.J. (2002). "Development of efficient threshold strategies for frequency doubling technology perimetry using computer simulation." Investigative Ophthalmology and Visual Science 43: 322-331.
Tyler, C.W. and Apkarian, P.A. (1985). "Effects of contrast, orientation and binocularity in the pattern evoked potential." Vision Research 25: 755-766.
243 Ungerleider, L.G. and Mishkin, M. (1982). Two cortical visual streams. In: Ingle, D.J. and Mansfield, R.J.W. (eds). Analysis of Visual Behavior. Massachusetts, MIT Press: pp 549-586.
Uren, S.M., Stewart, P. and Crosby, P.A. (1979). "Subject cooperation and the visual evoked response." Investigative Ophthalmology and Visual Science 18: 648-652.
Valberg, A., Olsen, B.T. and Marthinsen, S. (1981). "Peripheral contrast reversal inhibits visually evoked potentials in the fovea." Vision Research 21: 947-950.
Valeton, J.M. and Van Norren, D. (1979). "Retinal site of transient tritanopia." Nature 280: 488-490.
Valfre, W., Rainero, I., Bergui, M. and Pinessi, L. (2008). "Voxel-based morphometry reveals gray matter abnormalities in migraine." Headache 48: 109-117.
Vecchia, D. and Pietrobon, D. (2012). "Migraine: a disorder of brain excitatory-inhibitory balance?" Trends in Neurosciences 35: 507-520.
Vidyasagar, T.R. and Urbas, J.V. (1982). "Orientation sensitivity of cat LGN neurones with and without inputs from visual cortical areas 17 and 18." Experimental Brain Research 46: 157-169.
Vigano, A., D'Elia, T.S., Sava, S.L., Auve, M., De Pasqua, V., Colosimo, A., Di Piero, V., Schoenen, J. and Magis, D. (2013). "Transcranial Direct Current Stimulation (tDCS) of the visual cortex: a proof- of-concept study based on interictal electrophysiological abnormalities in migraine." The Journal of Headache and Pain 14: 23.
Vikelis, M. and Mitsikostas, D.D. (2007). "The role of glutamate and its receptors in migraine." Central Nervous System and Neurological Disorders. Drug Targets 6: 251-257.
Vincent, A.J., Spierings, E.L. and Messinger, H.B. (1989). "A controlled study of visual symptoms and eye strain factors in chronic headache." Headache 29: 523-527.
Vincent, M., Pedra, E., Mourao-Miranda, J., Bramati, I.E., Henrique, A.R. and Moll, J. (2003). "Enhanced interictal responsiveness of the migraineous visual cortex to incongruent bar stimulation: a functional MRI visual activation study." Cephalalgia 23: 860-868.
Vingrys, A., Demirel, S. and Kalloniatis, M. (1994). Multi-dimensional color, flicker and increment perimetry. In: Mills RP and Wall M (eds). Perimetry Update 1994/95. Proceedings of the XIth International Perimetric Society Meeting. Amsterdam/New York, Kugler Publications: pp 159-166.
Vingrys, A.J. and Helfrich, K.A. (1990). "The Opticom M-600: a new LED automated perimeter." Clinical and Experimental Optometry 73: 3-17.
Vingrys, A.J. and Pianta, M.J. (1998). "Developing a clinical probability density function for automated perimetry." Australian and New Zealand Journal of Ophthalmology 26: S101-103.
Vingrys, A.J. and Demirel, S. (1998). "False-response monitoring during automated perimetry." Optometry and Vision Science 75: 513-517.
Viswanathan, S., Frishman, L.J. and Robson, J.G. (2000). "The uniform field and pattern ERG in macaques with experimental glaucoma: removal of spiking activity." Investigative Ophthalmology and Visual Science 41: 2797-2810.
Wagner, D., Manahilov, V., Loffler, G., Gordon, G.E. and Dutton, G.N. (2010). "Visual noise selectively degrades vision in migraine." Investigative Ophthalmology and Visual Science 51: 2294-2299.
244 Wagner, D., Manahilov, V., Gordon, G.E. and Storch, P. (2012). "Long-range inhibitory mechanisms in the visual system are impaired in migraine sufferers." Cephalalgia 32: 1071-1075.
Wagner, D., Manahilov, V., Gordon, G.E. and Loffler, G. (2013). "Global shape processing deficits are amplified by temporal masking in migraine." Investigative Ophthalmology and Visual Science 54: 1160- 1168.
Wakakura, M. and Ichibe, Y. (1992). "Permanent homonymous hemianopias following migraine." Journal of Clinical Neuroophthalmology 12: 198-202.
Wall, M., Maw, R.J., Stanek, K.E. and Chauhan, B.C. (1996). "The psychometric function and reaction times of automated perimetry in normal and abnormal areas of the visual field in patients with glaucoma." Investigative Ophthalmology and Visual Science 37: 878-885.
Walsh, V. and Cowey, A. (1998). "Magnetic stimulation studies of visual cognition." Trends in Cognitive Sciences 2: 103-110.
Wang, J., Mitchell, P. and Smith, W. (1997). "Is there an association between migraine headache and open-angle glaucoma? Findings from the Blue Mountains Eye Study." Ophthalmology 104: 1714-1719.
Wang, W., Wang, G.P., Ding, X.L. and Wang, Y.H. (1999). "Personality and response to repeated visual stimulation in migraine and tension-type headaches." Cephalalgia 19: 718-724.
Watamaniuk, S.N., Sekuler, R. and Williams, D.W. (1989). "Direction perception in complex dynamic displays: the integration of direction information." Vision Research 29: 47-59.
Watanabe, M. and Rodieck, R.W. (1989). "Parasol and midget ganglion cells of the primate retina." The Journal of Comparative Neurology 289: 434-454.
Webb, B.S., Dhruv, N.T., Solomon, S.G., Tailby, C. and Lennie, P. (2005). "Early and late mechanisms of surround suppression in striate cortex of macaque." The Journal of Neuroscience 25: 11666-11675.
Weber, J. and Klimaschka, T. (1995). "Test time and efficiency of the dynamic strategy in glaucoma perimetry." German Journal of Ophthalmology 4: 25-31.
Webster, K.E., Edwin Dickinson, J., Battista, J., McKendrick, A.M. and Badcock, D.R. (2011). "Increased internal noise cannot account for motion coherence processing deficits in migraine." Cephalalgia 31: 1199-1210.
Webster, K.E., Dickinson, J.E., Battista, J., McKendrick, A.M. and Badcock, D.R. (2012). "Evidence for increased internal noise in migraineurs for contrast and shape processing." Cephalalgia 32: 125-139.
Weijland, A., Fankhauser, F., Bebie, H. and Flammer, J. (2004). Automated Perimetry: Visual Field Digest. Köniz, Haag-Streit AG.
Weiller, C., May, A., Limmroth, V., Juptner, M., Kaube, H., Schayck, R.V., Coenen, H.H. and Diener, H.C. (1995). "Brain stem activation in spontaneous human migraine attacks." Nature Medicine 1: 658- 660.
Weisinger, H., Vingrys, A. and Sinclair, A. (1996). "Electrodiagnostic methods in vision. Part 1. Clinical application and measurement." Clinical and Experimental Optometry 79: 50-61.
Welch, K.M., D'Andrea, G., Tepley, N., Barkley, G. and Ramadan, N.M. (1990). "The concept of migraine as a state of central neuronal hyperexcitability." Neurologic Clinics 8: 817-828.
245 Wild, J.M., Cubbidge, R.P., Pacey, I.E. and Robinson, R. (1998). "Statistical aspects of the normal visual field in short-wavelength automated perimetry." Investigative Ophthalmology and Visual Science 39: 54-63.
Wilkins, A., Nimmo-Smith, I., Tait, A., McManus, C., Della Sala, S., Tilley, A., Arnold, K., Barrie, M. and Scott, S. (1984). "A neurological basis for visual discomfort." Brain 107: 989-1017.
Wilkinson, F., Wilson, H.R. and Habak, C. (1998). "Detection and recognition of radial frequency patterns." Vision Research 38: 3555-3568.
Wilkinson, F., James, T.W., Wilson, H.R., Gati, J.S., Menon, R.S. and Goodale, M.A. (2000). "An fMRI study of the selective activation of human extrastriate form vision areas by radial and concentric gratings." Current Biology 10: 1455-1458.
Wilkinson, F. and Crotogino, J. (2000). "Orientation discrimination thresholds in migraine: a measure of visual cortical inhibition." Cephalalgia 20: 57-66.
Wilkinson, F. (2004). "Auras and other hallucinations: windows on the visual brain." Progress in Brain Research 144: 305-320.
Wilkinson, F., Karanovic, O. and Wilson, H.R. (2008). "Binocular rivalry in migraine." Cephalalgia 28: 1327-1338.
Wilson, H.R. and Humanski, R. (1993). "Spatial frequency adaptation and contrast gain control." Vision Research 33: 1133-1149.
Wilson, H.R. and Wilkinson, F. (1998). "Detection of global structure in Glass patterns: implications for form vision." Vision Research 38: 2933-2947.
Wilson, H.R. and Kim, J. (1998). "Dynamics of a divisive gain control in human vision." Vision Research 38: 2735-2741.
Wilson, H.R. (2007). "Minimal physiological conditions for binocular rivalry and rivalry memory." Vision Research 47: 2741-2750.
Wilson, N.R., Runyan, C.A., Wang, F.L. and Sur, M. (2012). "Division and subtraction by distinct cortical inhibitory networks in vivo." Nature 488: 343-348.
Wolff, H. (1963). Headache and Other Head Pain. New York, Oxford University Press.
Wollstein, G., Garway-Heath, D.F. and Hitchings, R.A. (1998). "Identification of early glaucoma cases with the scanning laser ophthalmoscope." Ophthalmology 105: 1557-1563.
Wolthausen, J., Sternberg, S., Gerloff, C. and May, A. (2009). "Are cortical spreading depression and headache in migraine causally linked?" Cephalalgia 29: 244-249.
Woods, R.P., Iacoboni, M. and Mazziotta, J.C. (1994). "Bilateral spreading cerebral hypoperfusion during spontaneous migraine headache." New England Journal of Medicine 331: 1689-1692.
World Health Organisation (2001). The World Health Report 2001: Mental health, new understanding, new hope. Geneva, Switzerland.
Xing, J. and Heeger, D.J. (2001). "Measurement and modeling of center-surround suppression and enhancement." Vision Research 41: 571-583.
246 Yenice, O., Temel, A., Incili, B. and Tuncer, N. (2006). "Short-wavelength automated perimetry in patients with migraine." Graefes Archives of Clinical and Experimental Ophthalmology 244: 589-595.
Yenice, O., Onal, S., Incili, B., Temel, A., Afsar, N. and Tanridag, T. (2007). "Assessment of spatial- contrast function and short-wavelength sensitivity deficits in patients with migraine." Eye 21: 218-223.
Yoon, J.H., Maddock, R.J., Rokem, A., Silver, M.A., Minzenberg, M.J., Ragland, J.D. and Carter, C.S. (2010). "GABA concentration is reduced in visual cortex in schizophrenia and correlates with orientation-specific surround suppression." The Journal of Neuroscience 30: 3777-3781.
Yu, C., Klein, S.A. and Levi, D.M. (2001). "Surround modulation of perceived contrast and the role of brightness induction." Journal of Vision 1: 18-31.
Zahavi, I., Chagnac, A., Hering, R., Davidovich, S. and Kuritzky, A. (1984). "Prevalence of Raynaud's phenomenon in patients with migraine." Archives of Internal Medicine 144: 742-744.
Zaletel, M., Strucl, M., Bajrovic, F.F. and Pogacnik, T. (2005). "Coupling between visual evoked cerebral blood flow velocity responses and visual evoked potentials in migraneurs." Cephalalgia 25: 567-574.
Zapf, H.R. and Bach, M. (1999). "The contrast characteristic of the pattern electroretinogram depends on temporal frequency." Graefes Archives of Clinical and Experimental Ophthalmology 237: 93-99.
Zemon, V., Kaplan, E. and Ratliff, F. (1980). "Bicuculline enhances a negative component and diminishes a positive component of the visual evoked cortical potential in the cat." Proceedings of the National Academy of Sciences of the United States of America 77: 7476-7478.
Zemon, V. and Ratliff, F. (1982). "Visual evoked potentials: evidence for lateral interactions." Proceedings of the National Academy of Sciences of the United States of America 79: 5723-5726.
Zenger-Landolt, B. and Heeger, D.J. (2003). "Response suppression in V1 agrees with psychophysics of surround masking." The Journal of Neuroscience 23: 6884-6893.
247
248 7 | Appendix A
This appendix includes supplementary material to Experiment 1 (Chapter 2).
7.1 Participants
7.1.1 Medication usage
At the onset or during a migraine attack, the migraine participants of Experiment 1 (Chapter 2) reported the use of the medications listed in Table 7.1 for pain relief and anti-nausea. Some participants reported the use of multiple medications.
Table 7.1 – List of medications used by migraine participants in Experiment 1 (n = 45).
Type of medication Medications Number of migraine participants
Analgesic Naproxon/mefenamic acid (NSAID) 1
Panadol (paracetemol) 22
Nurofen/Advil (ibuprofen) 19
Aspirin (acetylsalicylic acid) 7
Mersyndol/Panadeine 8
(paracetemol + codeine)
Codeine (morphine) 2
Anti-migraine preparations Triptans 6
Ergotamines 1
Antiemetic (anti-nausea) Maxolon (metoclopramide) 1
None 4
249 7.1.2 Questionnaires
The Headache Questionnaire was used to verify headache status and aid in the classification of control and migraine participants. The Migraine Disability Assessment Score (MIDAS) questionnaire was administered to all migraine participants to describe the impact of their attacks on activities of daily living.
7.2 Calibration of equipment
7.2.1.1 Gamma corrections
The CRT monitor used for electrophysiology (Sony G520 Trinitron, 1024 × 768 pixels, 100 Hz frame rate) was calibrated using a ColorCal colorimeter with photopic filter (Cambridge Research Systems, Kent, UK) approximately monthly over the course of 3.5 years. The CRT comprised three colour phosphor guns: red (1931 CIE chromaticity co-ordinates: x = 0.64, y = 0.35), green (x = 0.29, y = 0.60), and blue (x = 0.15, y = 0.07). The Michelson contrast (Equation 2.1, see Chapter 2) of the checkerboard stimulus was calibrated approximately monthly (averaged across four readings) for input contrasts ranging from 0% (uniform grey screen) to 100% in 5% contrast steps.
250
251
252
Figure 7.1 – Mean output contrast as a function of input contrast, measured in 5% contrast steps. Error bars represent the standard deviation. The maximum contrast that could be generated over the course of 3.5 years was 96 ± 0.6%. The solid line indicates the line-of-best-fit through the mean data (R2 = 0.99, slope = 0.98 ± 0.003, p < 0.001).
7.2.2 Mean luminance
Over the course of 3.5 years, the maximum luminance of the CRT monitor decreased with time (Figure 7.2A, solid line, R2 = 0.82, slope: p < 0.001). Despite this, the mean luminance at the centre of the screen during pattern-reversal stimulation at 1 Hz and 8.3 Hz remained constant (Figure 7.2B). There was no significant change in mean luminance over time (R2 = 0.02, slope: p = 0.46). The mean luminance did not fall below 50 cd/m2 and was, on average, 53 cd/m2. Because CRT monitors do not provide uniform output across the screen, mean luminance was also measured at the four corners of the screen. Peripheral mean luminance did not fall below 90% of the centre (Figure 7.3).
253
Figure 7.2 – Luminance of the CRT monitor over the course of 3.5 years. (A) Maximum luminance (B) Mean luminance during pattern-reversal stimulation. The solid line indicates the line-of-best-fit for maximum luminance (R2 = 0.82, slope = -0.008 ± 0.0006, p < 0.001). Horizontal dotted lines are the 5th and 95th percentiles of mean luminance.
254
Figure 7.3 – Uniformity of mean luminance across the screen over the course of 3.5 years. Measurements were taken at the (A) top left of screen (B) top right of screen (C) bottom left of screen (D) bottom right of screen. Mean luminance was expressed as a percentage of the luminance measured at the centre of the screen. Horizontal dotted lines indicate where peripheral mean luminance was identical to the centre (100%).
7.2.3 Temporal frequency
The temporal frequency of the stimulus used for steady-state recordings was verified using a photodiode connected to a data acquisition board with digitisation rate of 100 kHz (National Instruments, Texas, USA). At 8.3 Hz, the checkerboard reversed every 60 ms; hence, one complete cycle from light (ON) to dark (OFF) occurred every 120 ms (Figure 7.4).
255
Figure 7.4 – Temporal characteristics of the steady-state stimulus. The relative change in voltage over time was proportional to the change in luminance of a single check. The voltage spikes were perceived as continuous (dotted line) and not as individual pulses of light, as their frequency was above the critical flicker fusion frequency. A complete cycle from dark (OFF, zero voltage) to light (ON, maximum voltage) occurred over 120 ms, corresponding to a reversal rate of 16.7 reversals/second (8.3 Hz).
7.3 Running average filter
In order to remove the high-frequency oscillations (noise) in the transient response waveforms, a centred, zero-stuffed running average (sample window: 21 ms) was applied using Microsoft Excel (Microsoft, Redmond, WA, USA). The running average acts as a low-pass filter. Figure 7.5 shows that the filter was capable of reducing the power of high frequencies, e.g. 50 Hz mains interference, resulting in a smoother waveform. The frequency response of the running average filter is depicted in Figure 7.6, revealing a step response with effective cut-off frequency (-3 dB) at 29 Hz.
256
Figure 7.5 – The running average as a filter for high frequency noise. (A) Example raw transient PERG waveform (B) Filtered waveform after application of a centred, running average (21 ms sample window).
Figure 7.6 – Frequency response of the moving average filter. Dashed lines indicate effective cut-off frequency (-3 dB) of moving average at 29 Hz.
257 7.4 Representative waveforms
The following figures depict the raw waveforms and corresponding Fourier spectra, where applicable, that span the control, MO, and MA groups (25th, 50th, and 75th percentiles) for the transient and steady-state PERG and PVER.
Figure 7.7 – Representative waveforms for Experiment 1, transient PVER. The 25th, 50th, and 75th percentiles of P100 amplitude are shown for each group (control, MO, MA).
258
Figure 7.8 – Representative waveforms and corresponding Fourier spectra for Experiment 1, steady-state PVER. The 25th, 50th, and 75th percentiles of 2F amplitude are shown for each group (control, MO, MA).
259
Figure 7.9 – Representative waveforms for Experiment 1, transient PERG. The 25th, 50th, and 75th percentiles of P50 amplitude are shown for each group (control, MO, MA).
260
Figure 7.10 – Representative waveforms and corresponding Fourier spectra for Experiment 1, steady-state PERG. The 25th, 50th, and 75th percentiles of 2F amplitude are shown for each group (control, MO, MA).
261 7.5 Extra statistical analyses
7.5.1 Pupil sizes
Prior to electrophysiological recordings, pupil sizes were measured using a ruler at the same ambient light levels and at the 50 cm working distance as the actual recordings. There was no difference in natural pupil size (ranging from 5 to 6.5 mm) between right and left eyes (RM-ANOVA main effect of eye: F (1, 73) = 1.51, p = 0.22) and between control and migraine participants (RM-ANOVA main effect of group: F (1, 73) = 0.03, p = 0.85).
7.5.2 Right versus left eyes
The tables below provide justification for presenting only the right eye data in Experiment 1 (Chapter 2), as ‘eye’ was not identified as a significant within-factor contributing to between-group differences for all PERG and PVER measures.
262 Table 7.2 – RM-ANCOVA for group comparisons of PVER amplitude. Age was the co- variate. Between factor was group (control, MO, MA) and within factors were eye (right, left) and component (P100, N135, 2F amplitude). DF: Degrees of freedom; SS: Sum of squares; MS: Mean square.
DF (error) SS MS F p-value
Group 2 (70) 480.11 240.05 3.59 0.03
Age 1 (70) 10.28 10.28 0.15 0.70
Eye 1 (70) 0.22 0.22 0.37 0.55
Eye x group 2 (70) 0.97 0.49 0.83 0.44
Eye x age 1 (70) 0.05 0.05 0.09 0.77
Eye x component a 1.47 (102.78) 0.13 0.09 0.06 0.90
Eye x component x group a 2.94 (102.78) 2.81 0.96 0.61 0.61
Eye x component x age a 1.47 (102.78) 0.24 0.16 0.10 0.84
Component b 1.87 (130.82) 192.30 102.89 4.29 0.02
Component x group b 3.74 (130.82) 325.08 86.97 3.63 0.009
Component x age b 1.87 (130.82) 129.26 69.16 2.89 0.06 a Mauchly’s test of sphericity: χ2(2)=40.06, p<0.001; Huynh-Feldt correction to degrees of freedom: ε=0.73 b Mauchly’s test of sphericity: χ2(2)=10.60 p=0.005; Huynh-Feldt correction to degrees of freedom: ε=0.93
263 Table 7.3 – RM-ANCOVA for group comparisons of PVER timing. Age was a co-variate. Between factor was group (control, MO, MA) and within factors were eye (right, left) and component (N75, P100, N135 peak times and 2F phase). DF: Degrees of freedom; SS: Sum of squares; MS: Mean square.
DF (error) SS MS F p-value
Group 2 (70) 157.90 78.95 0.60 0.55
Age 1 (70) 11.75 11.75 0.09 0.77
Eye 1 (70) 1.44 1.44 1.09 0.30
Eye x group 2 (70) 1.59 0.79 0.60 0.55
Eye x age 1 (70) 0.34 0.34 0.26 0.61
Eye x component a 1.54 (107.42) 7.14 4.65 0.64 0.49
Eye x component x group a 3.07 (107.42) 29.32 9.55 1.32 0.27
Eye x component x age a 1.54 (107.42) 2.68 1.75 0.24 0.73
Component b 1.61 (112.40) 71679.51 44642.20 314.56 < 0.001
Component x group b 3.21 (112.40) 162.11 50.48 0.36 0.80
Component x age b 1.61 (112.40) 66.45 41.38 0.29 0.70 a Mauchly’s test of sphericity: χ2(5)=147.12, p<0.001; Huynh-Feldt correction to degrees of freedom: ε=0.51 b Mauchly’s test of sphericity: χ2(5)=112.50, p<0.001; Huynh-Feldt correction to degrees of freedom: ε=0.54
264 Table 7.4 – RM-ANCOVA for group comparisons of PERG amplitude. Age was the co- variate. Between factor was group (control, MO, MA) and within factors were eye (right, left) and component (P50, N95, 2F amplitude). DF: Degrees of freedom; SS: Sum of squares; MS: Mean square.
DF (error) SS MS F p-value
Group 2 (71) 10.79 5.40 0.36 0.70
Age 1 (71) 17.36 17.36 1.15 0.29
Eye 1 (71) 0.29 0.29 0.97 0.33
Eye x group 2 (71) 0.28 0.14 0.48 0.62
Eye x age 1 (71) 0.19 0.19 0.66 0.42
Eye x component a 1.29 (91.71) 2.96 2.29 2.18 0.14
Eye x component x group a 2.58 (91.71) 1.52 0.59 0.56 0.62
Eye x component x age a 1.29 (91.71) 1.56 1.21 1.15 0.30
Component b 1.53 (108.94) 238.16 155.22 39.29 < 0.001
Component x group b 3.07 (108.94) 6.33 2.06 0.52 0.67
Component x age b 1.53 (108.94) 7.79 5.08 1.29 0.28 a Mauchly’s test of sphericity: χ2(2)=69.26, p<0.001; Huynh-Feldt correction to degrees of freedom: ε=0.65 b Mauchly’s test of sphericity: χ2(2)=33.44, p<0.001; Huynh-Feldt correction to degrees of freedom: ε=0.77
265 Table 7.5 – RM-ANCOVA for group comparisons of PERG timing. Age was the co- variate. Between factor was group (control, MO, MA) and within factors were eye (right, left) and component (N35, P50, N95 peak times and 2F phase). DF: Degrees of freedom; SS: Sum of squares; MS: Mean square.
DF (error) SS MS F p-value
Group 2 (71) 59.50 29.75 1.01 0.37
Age 1 (71) 212.58 212.58 7.19 0.009
Eye 1 (71) 0.61 0.61 0.78 0.38
Eye x group 2 (71) 2.00 2.00 1.29 0.28
Eye x age 1 (71) 0.53 0.53 0.69 0.41
Eye x component a 2.04 (144.81) 2.46 1.21 0.84 0.44
Eye x component x group a 4.08 (144.81) 10.01 2.45 1.71 0.15
Eye x component x age a 2.04 (144.81) 6.23 3.06 2.13 0.12
Component b 1.70 (120.90) 36521.07 21446.99 912.57 < 0.001
Component x group b 3.41 (120.90) 70.18 20.61 0.88 0.47
Component x age b 1.70 (120.90) 94.65 55.58 2.37 0.11 a Mauchly’s test of sphericity: χ2(5)=100.33, p<0.001; Huynh-Feldt correction to degrees of freedom: ε=0.68 b Mauchly’s test of sphericity: χ2(5)=105.12, p<0.001; Huynh-Feldt correction to degrees of freedom: ε=0.57
266 7.5.3 Exclusion of pre-attack data
To avoid the possible confound of altered PVER amplitudes measured in the pre- attack period, a second analysis was conducted where data from the four individuals who experienced a migraine one day after testing were excluded. As demonstrated by the tables below, group comparisons were not different from the results of Experiment 1 (Chapter 2), which justifies the inclusion of these participants in the original analysis.
Table 7.6 – RM-ANCOVA for group comparisons of PVER amplitude, excluding data from the four participants who were tested one day before a migraine. Age was the co- variate. Between factor was group (control, MO, MA) and within factors were eye (right, left) and component (P100, N135, 2F amplitude). DF: Degrees of freedom; SS: Sum of squares; MS: Mean square.
DF (error) SS MS F p-value
Group 2 (66) 488.87 244.43 3.57 0.03
Group x component a 3.79 (125.1) 311.49 82.14 3.38 0.01 a Mauchly’s test of sphericity: χ2(2)=9.12, p=0.01; Huynh-Feldt correction: ε=0.95
Table 7.7 – Summary of PVER parameters, excluding data from the four participants who were tested one day before a migraine (mean ± standard deviation).
Parameter Controls MO MA p-valuea
P100 amplitude (µV) 11.9 ± 5.2 12.7 ± 5.7 8.1 ± 3.4 ** 0.02
N135 amplitude (µV) 13.4 ± 5.8 16.5 ± 6.5 12.1 ± 4.7 0.04
2F amplitude (µV) 3.7 ± 1.9 2.6 ± 1.5 ** 2.2 ± 1.4 ** 0.003 a p-values represent the main effect of group (RM-ANCOVA) ** denotes significant difference from controls (post-hoc pairwise Bonferroni tests between groups, p < 0.05)
267 7.5.4 Responses to smaller elements
ISCEV guidelines for PVER recommend that the response to smaller sized checks (0.25°) is measured in conjunction with the standard 0.8° checks. The normal response to 0.25° checks outlined in Tables 7.8 and 7.9 confirms the absence of optical problems that may affect the response to higher spatial frequencies.
Table 7.8 – Summary of transient PVER parameters measured in response to 0.25° checks (mean ± standard deviation).
Parameter Controls MO MA p-valuea
P100 amplitude (µV) 12.4 ± 5.8 12.9 ± 5.6 9.7 ± 3.8 0.13
N135 amplitude (µV) 13.0 ± 6.0 16.5 ± 8.1 12.9 ± 6.1 0.10
N75 time (ms) 80 ± 4 80 ± 4 79 ± 4 0.30
P100 time (ms) 106 ± 6 107 ± 5 106 ± 5 0.79
N135 time (ms) 153 ± 13 155 ± 12 154 ± 19 0.85 a p-values represent the main effect of group (RM-ANCOVA)
Table 7.9 – Summary of transient PERG parameters measured in response to 0.25° checks (mean ± standard deviation).
Parameter Controls MO MA p-valuea
P50 amplitude (µV) 5.0 ± 1.6 4.8 ± 1.7 5.5 ± 1.8 0.64
N95 amplitude (µV) 7.8 ± 2.1 7.3 ± 2.3 8.2 ± 2.7 0.61
N35 time (ms) 25 ± 3 25 ± 4 25 ± 4 0.63
P50 time (ms) 56 ± 3 55 ± 3 56 ± 3 0.28
N95 time (ms) 109 ± 5 107 ± 6 108 ± 5 0.41 a p-values represent the main effect of group (RM-ANCOVA)
7.6 Published work
The following journal article has been published from the work arising from Experiment 1 (Chapter 2).
268 Original Article
Cephalalgia 32(12) 896–907 ! International Headache Society 2012 Simultaneous retinal and cortical visually Reprints and permissions: sagepub.co.uk/journalsPermissions.nav evoked electrophysiological responses in DOI: 10.1177/0333102412453953 between migraine attacks cep.sagepub.com
Bao N Nguyen, Allison M McKendrick and Algis J Vingrys
Abstract Purpose: People with migraine often report aversion to flickering lights and show abnormal results on behavioural tasks that require the processing of temporal visual information. Studies have reported that the cortically evoked electro- physiological response to a flickering visual stimulus is abnormal; however, none have considered whether there is an underlying pre-cortical abnormality. In this cross-sectional study, we consider whether people with migraine have retinal and cortical electrophysiological abnormalities to flickering stimuli. Methods: Monocular transient (1 Hz) and steady-state (8.3 Hz) pattern reversal electroretinograms (PERGs) and pattern visual evoked responses (PVERs) were measured simultaneously in 45 people with migraine (26 without aura, 19 with aura) and 30 non-headache controls at a time between migraine attacks. Results: PERG amplitude and timing did not differ significantly between groups. Transient PVER amplitude was significantly reduced (28%) in the migraine with aura group compared to the controls F(2,72) 3.6, p 0.03). Both migraine groups showed significant reductions (32%, 39%) in steady-state PVER amplitude relative¼ to controls¼ (F(2,70) 4.3, p 0.02). Conclusions: This study finds normal retinal processing of flickering stimuli in the presence of abnormal¼ cortical¼ function between migraine attacks.
Keywords Migraine, temporal processing, contrast processing, visual evoked potentials, electroretinogram
Date received: 2 April 2012; revised: 30 May 2012; accepted: 14 June 2012
Introduction Visual field deficits associated with migraine can be Most studies of visual processing in people with non-homonymous and unilateral (13–19) and some- migraine focus on cortical processing, as migraine is a times described as ‘glaucomatous’ because of their loca- disorder of the central nervous system (1). Abnormal lised and arcuate pattern (13,17). It has been suggested performance on cortically processed behavioural tasks that such deficits may arise from peripheral vascular has been well documented between migraine attacks causes (17,20), with indirect evidence being an associ- (2–8). Such tasks rely on normal contrast processing, ation between migraine and peripheral vasospastic ten- which begins in the retina, then the lateral geniculate dencies (20), including transient retinal blood flow nucleus, before further processing in the primary visual changes (21). Moreover, migraine is a recognised risk cortex (V1) (9). It is important to ensure that differences attributed to cortical processing do not reflect reduced inputs from earlier, pre-cortical stages of the visual Department of Optometry and Vision Sciences, The University of pathway. Melbourne, Australia Several lines of research suggest non-cortical dys- Corresponding author: function in migraine. Between attacks, people with Allison M McKendrick, Department of Optometry and Vision Sciences, migraine show abnormal performance on behavioural The University of Melbourne, Parkville VIC 3010 Australia. tasks designed to assess pre-cortical function (8,10–12). Email: [email protected]
269 Nguyen et al. 897
factor for visual field progression in normal tension glaucoma (22) – a form of glaucoma associated with Methods a vascular aetiology (20) – and has been associated Participants with retinal nerve fibre layer loss (23). One method of comparing retinal to cortical func- Participants in this cross-sectional study were recruited tion in the same individual is by concurrent electro- via an advertisement within the University of physiological measurement of neuronal activity in the Melbourne community, as well as from a database of eye and brain. Pattern electrophysiology invokes activ- previous study participants, from June 2010 to ity of contrast-dependent neurones. The pattern elec- September 2011. Thirty non-headache controls (aged troretinogram (PERG) assesses the integrity of retinal 19–46 years, mean standard deviation [SD]: 26 7), ganglion cells, whereas the function of the visual path- 26 people with migraineÆ without aura (MO, aged 20–41Æ way, including V1, can be assessed with the pattern years, mean SD: 28 6) and 19 people with migraine visual evoked response (PVER) (24). The patterned with aura (MA,Æ agedÆ 19–43 years, mean SD: 33 6) stimulus can be contrast-reversed at different temporal were tested at the Department of OptometryÆ and VisionÆ frequencies. The standard electrophysiological response Sciences at the University of Melbourne. On average, is a transient response (<3 Hz), where the response is the MA group was older than the control and MO complete before the next contrast reversal. Transient groups (F(2,72) 6.88, p 0.002). The proportion of recording allows visualisation of a single waveform males to females¼ was not significantly¼ different between characterised by identifiable components. At higher each group (controls 9:21, MO: 4:22, MA: 5:14). temporal frequencies ( 4 Hz), the successive waveforms All participants had best corrected visual acuity of overlap and a steady-state! response is generated. 6/7.5 or better and refractive errors of less than 5.00 Steady-state electrophysiological recording is presumed D sphere and 2.00 D astigmatism. ParticipantsÆ were to measure similar mechanisms as behavioural meas- free from systemicÀ disease and medications known to ures of temporal processing (flicker). Flickering and affect visual function or neurological state, including moving stimuli have consistently identified differences prophylactic migraine medications, and had normal between migraine and non-headache groups, (2,3,5– findings in a comprehensive eye examination (slitlamp 8,11,14,19,25,26) including tests of pre-cortical function biomicroscopy, applanation tonometry, ophthalmos- (10,16). copy and optic nerve head imaging). All participants Previous electrophysiological studies during the were classified as within normal limits using the interictal period in migraine have yielded varying out- Moorfield’s regression analysis (MRA) tool of the comes (27,28). Many studies report anomalous PVER Heidelberg retinal tomograph (Heidelberg Engineering, results, but most do not consider whether the retina is Heidelberg, Germany). The MRA has excellent specificity normal. Only two studies have noted that the PERG (>85%) for cross-sectional diagnoses of open-angle glau- was normal in the presence of abnormal cortical func- coma (34). tion, both measured using transient stimulation (29,30). Control participants had never experienced a However, despite studies that have found abnormal migraine and were free from regular headaches (less cortical responses to steady-state stimulation (31–33), than four in the past year). All migraine participants the steady-state retinal response has not previously met the criteria of the International Headache Society been measured in migraine. for either MA or MO (35) and were tested at least seven The purpose of this study was to measure both tran- days after a migraine attack to minimise residual effects sient and steady-state retinal and cortical responses in of medications and the possible influence of transient the same individuals with migraine on the same day. post-migraine fatigue or nausea. Participants completed This will bring together a range of techniques that have a custom headache questionnaire and the Migraine been used previously in isolation to build up a more Disability Assessment Score (MIDAS) to determine a complete picture of how the electrophysiological variety of headache characteristics, summarised in response is different between migraine attacks. We Figure 1. The MIDAS questionnaire is a validated tool hypothesise that a concomitant retinal and cortical that scores the impact of headaches on tasks of daily abnormality would suggest that the cortical involve- living over the past three months (36). MIDAS scores ment arises, at least partly, downstream from retinal are interpreted as minimal (grade 1, score 0–5), mild dysfunction. To our knowledge, this is the first study (grade 2, score 6–10), moderate (grade 3, score 11–20) to measure steady-state and transient electrophysio- or severe disability (grade 4, score 21 ). Six of the 30 logical responses at the retina and cortex simultan- control participants (20%) reported thatþ only one family eously, which will reveal whether the cortical member suffered from migraines (e.g. mother, sister, abnormalities are explained by aberrant function earlier grandmother), whereas 33 of the 45 participants with in the visual pathway. migraine (73%) had at least one other family member
270 898 Cephalalgia 32(12)
(A)40 (B)40 (C) 1500 p=0.06 p=0.01 p=0.46 1250 30 30 1000
20 20 750
500 years of migraine 10 10
age at first migraine (years) 250 estimated number of lifetime attacks 0 0 0
MO MA MO MA MO MA
(D)300 (E)30 (F) 100 p=0.31 p=0.12 p=0.01 250 25 80
200 20 60 150 15 40 100 10 MIDAS score (days)
days since last migraine 20 50 5
0 number of weeks between migraines 0 0
MO MA MO MA MO MA
Figure 1. Comparison of migraine characteristics between the MO and MA groups. The boxes represent the 25th, 50th and 75th percentiles, with the whiskers showing the 10th and 90th percentiles. All outliers are shown as individual symbols. (A) Age at first migraine, (B) years of migraine (estimated from the age of first migraine), (C) number of lifetime attacks (estimated from migraine frequency and years of migraine), (D) days since last migraine, (E) number of weeks between migraines (estimated from the self-reported number of migraines in past year), (F) MIDAS score, the total number of days of lost productivity due to migraine over the past three months. MO: migraine without aura; MA: migraine with aura; MIDAS: Migraine Disability Assessment Score. who suffered from migraines (e.g. parent, sibling, grand- All participants provided written informed consent parent, aunt, cousin). prior to testing in accordance with a protocol approved Of the eighty people who agreed to participate in the by the Human Research Ethics Committee of the study, five participants attended the test session but did University of Melbourne and compliant with the not complete the testing, as they were excluded for the tenets of the Declaration of Helsinki. following reasons: ptosis (one control), glaucoma sus- pect (one MO), more than four headaches in the past Timing of the test visit year (four controls) and high intraocular pressure (one control). Thus, the final sample size for analysis was 75 For migraine participants, the test visit was scheduled participants (30 non-headache, 26 MO, 19 MA). at least seven days after a migraine. Albeit not a formal A power analysis was performed using data from studies requirement of our study, all participants were asked to that have previously used the steady-state PERG to test advise of any adverse effects in the days after the test people with early glaucoma (37,38). The power analysis session, including when their next migraine occurred. In indicated that 17 participants in each group provided a the majority of cases, participants responded to our power of 80% for detecting a reduction (alpha of 0.05) in follow-up attempts that they did not have another the steady-state PERG response of one-third the mag- migraine in the 72 hours after the test. Of the 45 nitude observed in people with early glaucoma, because people with migraine tested, four participants reported more subtle deficits are expected with migraine. a migraine the day after testing. Data from these
271 Nguyen et al. 899
ground (A) inactive PVER electrode electrode 20%
40% Cz 30% Fz active PVER electrode Oz nasion inactive PERG 10% electrode inion
active (B) PERG electrode OL OZ OR
10%
Figure 2. Electrode locations for simultaneous standard PVER/PERG recording. (A) For the PVER, scalp electrodes are placed in proportion to the size of the head in accordance with the International 10-20 system, which divides the distance between the two bony landmarks, the nasion and inion, into 10% and 20% proportions. The active PVER electrode is located along the vertical midline at OZ and the inactive reference electrode at FZ. For the PERG, the active electrode is a DTL carbon-fibre electrode placed along the lower eyelid margin. The inactive PERG electrode is a skin electrode placed at the cheekbone (left eye shown only). The common ground electrode is placed at CZ, midway between the nasion and inion on the midline. (B) The right and left lateral PVER electrodes are placed at OR and OL, respectively. PVER: pattern visual evoked response; PERG: pattern reversal electroretinogram.
participants are represented as unfilled symbols (see A 0.5-diameter red square provided a central fixation Figure 4). target. PERG responses were recorded with corneal DTL Pattern electrophysiology electrodes near the lower limbus and referenced to the ipsilateral canthus with silver-silver chloride electrodes. The PERG and PVER were recorded simultaneously The PVER was recorded with gold cup electrodes at Oz, from each eye according to ISCEV (International OR and OL, and referenced to Fz with the common Society for Clinical Electrophysiology of Vision) stand- ground at Cz (Figure 2). Electrode impedance was ards (39,40) using the Espion system (Diagnosys LLC, kept below 5 k . Signals were amplified, bandpass-fil- Cambridge, UK). A black-and-white square-wave tered (1.25–100 Hz) and digitised (1000 Hz) to 16-bit checkerboard stimulus was presented on a gamma-cor- resolution. Blink artefacts where signals exceeded rected Sony G520 21-inch CRT monitor (100 Hz, 100 mV were automatically rejected. As people with 1024 768 pixels). The checkerboard subtended 31 migraineÆ can find high-contrast patterns aversive (42), Â at a viewing distance of 50 cm, and consisted of 0.8 each stimulus presentation was limited to approxi- checks of 96% contrast and mean luminance of mately 20 seconds, corresponding to 25 sweeps, fol- 52 cd/m2. The contrast of the checkerboard was coun- lowed by a brief break. A total of 200 signals were terphased at 1 Hz (transient) and 8.3 Hz (steady-state). averaged for each stimulus condition of differing spa- In addition, the transient response to 0.25 checks was tiotemporal frequency. To confirm intrasessional repro- measured to confirm the absence of optical problems ducibility, we computed an index of response that may affect the response to finer checks (39). The consistency (39,40), the coefficient of variation (COV) steady-state response to 16 checks was recorded to between two consecutive partial averages of 100 enable determination of the PERG ratio (41)—a sensi- sweeps. Typically, the COV of amplitude for the tive measure of PERG dysfunction. The different spa- PERG (transient 9.1%, steady-state 8.1%) and PVER tiotemporal frequencies were presented in random (transient 10.7%, steady-state 9.8%) were comparable order and counterbalanced between participants. to those reported in the literature (data not shown)
272 900 Cephalalgia 32(12)
(A) Transient PVER 31°
P100 amplitude A
B
N75 N135 amplitude 50 cm
0 50 100 150 200 250
(B)Transient PERG (C) Steady state
P50 amplitude DFT 2F amplitude
N35
N95 amplitude
0 50 100 150 200 250 0 10 20 30 40 50 60 70 Time (ms) Frequency (Hz)
Figure 3. Schematic of PERG/PVER recording setup and representative waveforms. (A) The transient PVER waveform is charac- terised by a prominent positive peak at 100 ms (vertical dotted line). The P100 amplitude is measured as the peak-to-peak distance from the first negative trough at 75 ms (N75) to the top of the positive peak. The N135 amplitude is taken from the P100 to the second negative trough (vertical dashed line). (B) The transient PERG waveform is characterised by a prominent positive peak at 50 ms (P50, vertical dotted line) and a negative trough (N95, vertical dashed line). Peak-to-peak P50 and N95 amplitudes are taken from N35 to P50, and P50 to N95, respectively. (C) Steady-state waveforms are resampled to give 512 data points and converted into the frequency domain by discrete Fourier transformation in order to measure the amplitude at the second harmonic (2 F). Noise is derived from the average amplitudes at the two neighbouring frequencies, 14.6 and 18.8 Hz (vertical dotted lines). PVER: pattern visual evoked response; PERG: pattern reversal electroretinogram.
(43,44). A typical recording session lasted approxi- the contrast-reversal frequency, which occurs at twice mately 35 minutes. the stimulation frequency (second harmonic 2F, Analysis was performed offline using Microsoft 16.7 Hz). Steady-state responses were excluded if the Excel (Microsoft, Redmond, WA, USA). Peak-to- 2F amplitude was not significantly different from noise peak amplitudes and peak times for the transient (p > 0.05), where noise was estimated from the average PERG P50 and N95 and PVER P100 and N135 amplitude at the two neighbouring frequencies (14.6 were measured according to ISCEV standards and 18.8 Hz) (46). Phase was calculated from the arc- (39,40). Retinocortical time was calculated as the tangent of the ratio of the real-to-imaginary compo- interlatency period from the PERG P50 to PVER nents of the discrete Fourier analysis. Because there P100, as a measure of signal transmission time can be an infinite set of phases separated by 2p between retina and cortex (45). Steady-state wave- radians with the same arctangent, the value modulo forms were analysed in the frequency domain after was returned (p rads). The phase convention was such discrete Fourier transformation of resampled wave- that increasing phase corresponded to a delay in the forms into a binary series (Figure 3). Because every signal. PVER interhemispheric asymmetry was defined change from black to white represents a single change as the absolute percentage amplitude difference in contrast, the largest contrast-evoked response is at between the response at OR and OL (29) (Figure 2).
273 Nguyen et al. 901
Table 1. Summary of PERG parameters (mean SD).a Æ Parameter Controls MO MA p value
Transient P50 amplitude (mV) 6.57 1.83 6.60 2.08 7.17 2.16 0.46 Æ Æ Æ N95 amplitude (mV) 10.3 2.52 10.6 3.07 10.6 2.75 0.81 Æ Æ Æ P50 time (ms) 51 2 51 3 51 2 0.26 Æ Æ Æ N95 time (ms) 104 5 105 5 104 5 0.37 Æ Æ Æ Steady-state 2F amplitude (mV) 3.52 0.818 3.25 1.10 3.43 1.02 0.66 Æ Æ Æ 2F phase (p rads) 1.4 0.8 1.6 0.6 1.4 0.7 0.99 Æ Æ Æ PERG ratio 1.2 0.2 1.3 0.4 1.3 0.3 0.23 Æ Æ Æ ap values represent the main effect of group (repeated measures analysis of co-variance [RM-ANCOVA]). PERG: pattern reversal electroretinogram; SD: standard deviation; MO: migraine without aura; MA: migraine with aura.
(A) Transient PVER P100 amplitude(B) Steady-state PVER amplitude 30 12 p=0.03 p=0.01 p=0.02 25 10 p=0.02 V) V) m m 20 8
15 6
10 4 PVER amplitude ( PVER amplitude (
5 2
0 0 control MO MA control MO MA
Figure 4. Distributions of the (A) transient PVER P100 amplitude and (B) steady-state PVER 2F amplitude in the control, MO and MA groups. The four participants who happened to be tested the day before a migraine are shown as unfilled symbols. Error bars indicate the mean SEM. PVER: pattern visualÆ evoked response; SEM: standard error of the mean.
A PERG ratio was computed by dividing the steady- state amplitudes to 0.8 and 16 checks (41). Results Migraine characteristics Statistical analysis Figure 1 shows boxplots of the migraine characteristics. Statistical analysis was performed using SPSS version The MA group had a longer duration of migraine his- 17.0 (SPSS Inc., Chicago, IL, USA). Age was entered as tory (Mann-Whitney rank sum test, p 0.01), whereas a covariate in all the analyses. Comparisons between the MO group reported significantly greater¼ impact of control and migraine groups were performed using a their migraines on daily activities, as reflected in the repeated measures analysis of co-variance (RM- higher MIDAS questionnaire scores (Mann-Whitney ANCOVA). Comparisons between migraine groups rank sum test, p 0.01). were performed using student t-tests or non-parametric ¼ Mann-Whitney rank sum tests where the data was non- Was there evidence for retinal dysfunction in Gaussian (Kolmogorov-Smirnov test). A p < 0.05 was people with migraine? considered significant for all statistical evaluations. As there was no difference between the left and right eyes Table 1 is a summary of the PERG findings. An RM- for any measures, the right eye response is represented ANCOVA (within factors: eye and component; in all figures. between factor: group) was performed separately for
274 902 Cephalalgia 32(12)
Table 2. Summary of PVER timing parameters (mean SD).a Æ Parameter Controls MO MA p value
Transient N75 time (ms) 73 4 71 5 71 5 0.41 Æ Æ Æ P100 time (ms) 102 5 101 4 101 5 0.62 Æ Æ Æ N135 time (ms) 146 16 146 8 145 15 0.85 Æ Æ Æ Retinocortical time (ms) 51 4 51 4 50 5 0.91 Æ Æ Æ Steady-state 2F phase (p rads) 0.7 0.4 0.5 0.2 0.5 0.2 0.11 Æ Æ Æ ap values represent the main effect of group (repeated measures analysis of co-variance [RM-ANCOVA]). PVER: pattern visual evoked response; SD: standard deviation; MO: migraine without aura; MA: migraine with aura.
transient PERG amplitudes and peak times, where the Both migraine groups were similarly affected (post-hoc components of interest were the P50 and N95. There Bonferroni multiple comparisons MO: p 0.02, MA: was no significant difference between the groups for any p 0.01). When normalised to the mean¼ control 2F of the transient measures (p > 0.05). For the 2F ampli- amplitude,¼ the MO group and the MA group amplitude tude, phase and PERG ratio, separate RM-ANCOVA were reduced, on average, by 32% and 39%, respect- (within factor: eye; between factor: group) analyses ively. There was no significant difference in phase showed that the groups did not differ for any of the between groups (Table 2). steady-state measures (p > 0.05). Because an increased interhemispheric asymmetry in PVER has been reported previously in some migraine Was there evidence for cortical dysfunction in studies (29,47–50), we recorded the PVER at the right people with migraine? and left hemispheres, and calculated an absolute amp- litude asymmetry. For the steady-state PVER, the lat- Figure 4 shows the group mean amplitudes for the tran- eral responses of three controls and four MA sient PVER. An RM-ANCOVA (within factors: eye participants were excluded because at least one and component; between factor: group) showed a sig- response was not significantly different from noise. nificant difference between groups, where the compo- Separate RM-ANCOVA (within factor: eye; between nents of interest were the P100 and N135 factor: groups) analyses showed that there was no dif- (F(2,71) 4.21, p 0.02). Both the control and MO ference in percentage asymmetry for either transient groups had¼ significantly¼ higher P100 amplitudes than (controls: 16.2 11.4%; MO: 21.2 15.5%; MA: the MA group (post-hoc Bonferroni multiple compari- 22.8 19.6%) orÆ steady-state responsesÆ (controls: sons, p 0.02), but there was no difference between the 18.2 Æ 13.2%; MO: 24.4 17.3%; MA: 19.4 9.3%). control¼ and MO groups. The difference between the GivenÆ that the PVER amplitudeÆ can be alteredÆ in the groups was more apparent for the P100 (significant pre-attack period (within 72 hours of a migraine) (51), interaction between group and component: we excluded the data from the four individuals who F(2,72) 3.58, p 0.03). To calculate the relative per- reported a migraine the day after the test session centage¼ reduction¼ in amplitude, each individual’s P100 (unfilled symbols in Figure 4). The results of the statis- amplitude was normalised to the average control P100 tical analyses were unchanged. amplitude. The MA group amplitude was, on average, reduced by 28%. As a control for optical factors, we Did the steady-state response reveal greater measured the transient PVER to smaller (0.25) checks dysfunction than the transient response? as per ISCEV standards (39), for which the P100 amp- litude did not differ between groups (data not shown). To compare steady-state to transient responses between Table 2 is a summary of all of the timing measures, groups, we analysed the P100 and 2F PVER amplitudes which did not differ between groups (RM-ANCOVA in a combined RM-ANCOVA (within factors: eye and within factor: eye; between factor: group). test type; between factor: group), where test type was Figure 4 shows the group mean steady-state PVER transient or steady-state (Figure 4). The PVER ampli- amplitudes. The steady-state response of one MA par- tude differed between the groups (F(2,70) 3.53, ticipant was excluded from analysis, as the signal was p 0.04). Post-hoc Bonferroni multiple comparisons¼ not significantly different to noise. Relative to controls, revealed¼ that the PVER amplitudes of the MA group the steady-state PVER amplitudes of the migraine were reduced for both test types relative to the MO groups were reduced (RM-ANCOVA: within factor: (p 0.05) and control groups (p 0.01), whereas the eye; between factor: group; F(2,70) 4.293, p 0.02). PVER¼ amplitudes of the MO¼ and control groups ¼ ¼
275 Nguyen et al. 903
were similar. The interaction between test type (transi- abnormal cortical response is unclear, as the PVER ent vs. steady-state) and group approached significance reflects the patency of input from LGN to V1, and (F(2,70) 2.91, p 0.06). Effect sizes (Cohen’s d) were includes the input of lateral connections within V1, as determined¼ to enable¼ comparison of the magnitude of well as feedback connections to V1 from higher cortical PVER deficit across both transient and steady-state areas. Our findings are consistent with previous studies recording, taking into consideration measurement where the transient PERG was normal but the PVER variability. The effect size (d) represents the difference was abnormal (29,30). between the control and MA groups in numbers of A novel component of this study was the addition of standard deviations, which was calculated as steady-state recording to the standard clinical protocol of transient recording, given the vast number of behav- d = ioural studies that report abnormal visual function with ¼ð m À cÞ pooled flickering or moving stimuli between migraine attacks where (2,3,5–8,10,11,14,16,18,19,25,26). Steady-state PERG measures are more sensitive than transient measures in identifying early retinal ganglion cell dysfunction 2 2 pooled =2 (41). In our study, both transient and steady-state ¼ ½ð m þ c Þ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi PERG measures were normal. Given behavioural evi- dence for pre-cortical dysfunction (8,10–12), our find- and m and c are the MA and control group mean ings do not necessarily preclude the presence of some PVER amplitudes, and m and c are the standard devi- retinal involvement in migraine. Rather, the standard ations. The effect sizes for the magnitude of the reduc- clinical protocols employed in this study do not identify tion in PVER amplitude in the MA group were large abnormalities in people with migraine. This result yet similar for the two test types (steady-state response: might arise if the test is insensitive to the type of dys- d 0.92; transient response: d 0.86). function possibly present in migraine. Unlike glau- ¼ ¼ coma, migraine is not a predominant and global Relationship with migraine features retinal ganglion cell dysfunction, for which the PERG is a sensitive indicator (24). If pre-cortical dysfunction Steady-state PVER amplitudes were reduced in both in migraine is of a more subtle or localised nature, it migraine groups relative to controls. To explore the may not be detected by our standard, full-field PERG possibility that the reduced cortical response was paradigm. Further investigations may make use of related to a migraine feature other than the presence multifocal ERG and VER techniques that enable high or absence of aura, Spearman’s rank correlation coef- spatial resolution and have the potential to provide ficients were determined for the entire migraine cohort. more information about the presence of localised There were no significant correlations between PVER visual field deficits in people with migraine (13–19). amplitude and any of the reported migraine character- For the transient PVER, the only abnormal finding istics (age at first migraine, duration of migraine his- was a reduced P100 amplitude within the MA group. tory, estimated number of lifetime attacks, days since Most PVER studies report increased P100 amplitude or last migraine, weeks between migraines and days of lost no difference at all (27,28). There are difficulties recon- productivity in past three months). ciling electrophysiological studies in migraine due to vast differences in experimental protocols. If we com- Discussion pare our results to those obtained under similar stimu- lus conditions (1 Hz), one study reported a 17% This study aimed to determine whether people with reduction in P100 amplitude in a pooled migraine migraine have abnormal retinal and/or cortical electro- group (52), while another reported a 36% reduction physiological responses between attacks, and included in P100 amplitude in people with at least 30 years of both steady-state and transient protocols. The steady- migraines with aura (53). In contrast, other studies state response is the response to rapidly alternating sti- report increased amplitude in specific migraine groups muli, whereas the transient response is seen to low tem- (left-sided hemicrania (54), MO and short-duration poral frequencies. Note that the terminology ‘sustained’ MA of less than 10 years (30)) or normal performance and ‘transient’ are often used to mean the opposite, (32,49). For the steady-state PVER, we found reduced outside of electrophysiology. A concomitant PERG amplitudes in both migraine groups. In contrast, previ- and PVER abnormality would suggest that the abnor- ous studies have in general reported higher steady-state mal cortical response arises, at least partly, downstream amplitudes, although the test protocols varied consid- from retinal dysfunction. There was no evidence for erably (31–33). Shibata and colleagues observed a sig- such an effect within our data. The exact site of the nificantly increased second harmonic amplitude in both
276 904 Cephalalgia 32(12)
MO and MA groups relative to controls at 5 Hz but not enhanced centre-surround suppression in migraine 10 Hz (31). The difference was evident for the grating (2,3), leading to a greater reduction in the perceived pattern of 0.5 cyc/deg spatial frequency (equivalent to contrast of a drifting central grating patch in the pres- 1.4 checks) and not for the smaller patterns tested (2 ence of a higher contrast surround of the same spatio- cyc/deg 0.4). Marrelli et al. tested children with temporal frequency, phase and orientation. Our migraine¼ and found higher amplitudes for the first har- extended checkerboard stimulus may have induced a monic response only (32). The presence of a significant similar centre-surround perceptual effect, which, if first harmonic, however, may indicate technical prob- enhanced in our migraine group would result in lems (40), as the greatest response is expected to occur increased suppression of the central stimulus and a rela- at the contrast-reversal frequency (second harmonic). tively more depressed PVER response relative to The protocol used by Diener et al. (33) is most similar controls. to ours (1 checks at 8.33 Hz), but the signals were not The reduction in PVER might alternatively reflect analysed in the frequency domain and therefore are structural abnormalities or metabolic disadvantage, difficult to compare to our findings. possibly as a cumulative result of repetitive ischaemia One possible interpretation of our data is that with migraine attacks (53). The use of advanced neu- reduced PVER amplitudes reflect a low cortical pre- roimaging techniques to rule out cortical lesions asso- activation level (55), which has been proposed to arise ciated with migraine (1) falls outside the scope of this from lower levels of central neuromodulators (e.g. sero- study. An indirect way to infer whether repeated tonin (56)). Indirect evidence for a reduced pre-activa- migraine episodes are associated with cumulative tion level in migraine in visual electrophysiology studies injury is by correlating years of migraine or attack fre- is the initial reduction in PVER amplitude relative to quency with the PVER, which in this study was not controls in the first block of averaged responses, typic- significant (data not shown). However, retrospective ally obtained within the first minute of recording reporting may not be accurate, and migraine frequency (57–59) or when two paired-pulse responses are mea- can vary over a lifetime. To determine whether an sured in close succession (60). Parallel findings of abnormally reduced PVER is a result of repeated reduced somatosensory evoked early high-frequency migraine attacks requires longitudinal investigation. It oscillations also imply decreased cortical activation in is also unknown whether damage needs to be cumula- both MA and MO (61). With increasing stimulation, tive, as one migraine might cause lasting damage. the cortical response is augmented up to a maximum It is worth considering whether our results are (or ‘ceiling’), at which point the response is reduced explicable by non-visual mechanisms. Prolonged testing (habituation) in non-headache controls (for a review, can induce drowsiness, thereby reducing the VER amp- see Coppola and colleagues (62)). A low pre-activation litude (24). Although attention was not specifically con- state may be protective against cortical hyperexcitabil- trolled, we have no reason to suspect that the people ity, the presumed anomalous neural state in migraine, with migraine were more drowsy or fatigued than our by allowing for a large range of activity before the ‘ceil- controls. The examiner frequently asked participants to ing’ is reached (55). Deficient habituation is the most indicate whether the fixation target was clear, and the consistent electrophysiological finding across multiple stimuli were interleaved with regular breaks. sensory modalities in migraine (62) and is typically Participants were tested at least seven days after demonstrated using continuous stimulation for at migraine offset to minimise migraine-related fatigue. least two minutes. As we did not wish to study habitu- There is also no evidence for greater accommodative ation, the stimuli in our study were presented for a fatigue in the migraine group, as their responses to short time (approximately 20 seconds) with brief the finer checks (0.25) were no different to controls breaks between each presentation, which is typical of (data not shown). An alternate possibility is greater clinical PVER recordings. Most PVER studies that aversion to the checkerboard stimulus in the migraine have not specifically explored habituation fail to specify group, as has been previously reported (26,42). Greater whether the recordings are made continuously. Without aversion may decrease the ability to voluntarily attend breaks between presentations, the increased PVER to the target and therefore reduce the PVER amplitude amplitude often reported may result from the averaging (24). We did not formally measure aversion; however, of progressively larger amplitudes over time due to the examiner informally asked for participant feedback abnormal habituation in migraine (55). regarding the testing. None of the participants reported Alternatively, changes in PVER amplitude might discomfort during the recordings or needed to abort the reflect anomalies in the balance between intracortical testing due to heightened aversion or an impending inhibition and excitation, in particular manifesting as migraine attack. It is worth noting that our participants a difference in perceptual centre-surround suppression. self-selected to be involved in a study with extensive Recently, Battista and colleagues demonstrated visual testing; hence, people with strong aversion to
277 Nguyen et al. 905
particular visual stimuli may have been less likely to frequencies using standard clinical protocols and volunteer. Our participant sample may not be represen- could not be explained by impairments in the eye or tative of a broader clinical group of people with the transmission along the visual pathway. Further migraine, who may have more severe events, chronic research is required to ascertain the underlying mech- migraines, or be taking prophylactic medications. anism for reduced PVER amplitude to identify the cor- In this study, the MO and MA groups differed in responding physiological (structural, vascular and/or their transient cortical response. The lack of a finding metabolic) differences in the migraine brain. of reduced transient PVER amplitude in the MO group implies that their deficit might show temporal tuning, by only being present at higher temporal frequency Funding stimulation (Figure 4). This is consistent with the tem- This work was supported by the National Health and poral tuning found in one MO participant by Medical Research Council (grant number 509208) and McKendrick et al. using behavioural methods (16). Australian Research Council (grant number FT0990930) to However, this study also reported greater deficits in author AMM. flicker contrast processing to higher temporal frequen- cies for MA subjects. It is still unclear whether a difference in temporal processing exists, given the References non-significant interaction between group and temporal 1. Schwedt TJ and Dodick DW. Advanced neuroimaging of frequency in this study. migraine. Lancet Neurol 2009; 8: 560–568. Alternatively, the two migraine subtype groups dif- 2. Battista J, Badcock DR and McKendrick AM. Migraine fered in age; however, the effect of age on PVER amp- increases centre-surround suppression for drifting visual litude is insignificant across the age range of our stimuli. PLoS One 2011; 6: e18211. migraine population (19–43 years) (63). Furthermore, 3. Battista J, Badcock DR and McKendrick AM. Center- age was not identified as a significant co-factor contri- surround visual motion processing in migraine. Invest buting to a between group difference for any of our Ophthalmol Vis Sci 2010; 51: 6070–6076. statistical analyses. The MA group also reported a 4. Tibber MS, Guedes A and Shepherd AJ. Orientation dis- longer migraine history. A previous study of 47 MA crimination and contrast detection thresholds in migraine for cardinal and oblique angles. Invest Ophthalmol Vis Sci and 37 MO participants reported a significant correl- 2006; 47: 5599–5604. ation between disease duration and reduced PVER 5. Ditchfield JA, McKendrick AM and Badcock DR. Pro- amplitude (53); however, that study comprised 14 par- cessing of global form and motion in migraineurs. Vision ticipants (15%) with at least a 30-year history of MA. Res 2006; 46: 141–148. Only one MA participant in our study had experienced 6. Antal A, Temme J, Nitsche MA, et al. Altered motion more than 30 years of migraine, so similar correlations perception in migraineurs: evidence for interictal cortical cannot be performed. People who experience aura are hyperexcitability. Cephalalgia 2005; 25: 788–794. at slightly greater risk of subclinical brain lesion, 7. McKendrick AM and Badcock DR. Motion processing stroke, cardiovascular disease and other ischaemic vas- deficits in migraine. Cephalalgia 2004; 24: 363–372. cular events than are those who experience MO (for a 8. McKendrick AM, Vingrys AJ, Badcock DR, et al. Visual dysfunction between migraine events. Invest Ophthalmol review, see Schwedt and Dodick (1) and Bigal and Vis Sci 2001; 42: 626–633. Lipton (64)). This suggests that the presence of aura 9. Kaplan E, Lee BB and Shapley R. New views of pri- may influence a person’s susceptibility to adverse vas- mate retinal function. In: Osborne N and Chader G cular effects, although the absolute risk is only margin- (eds) Progress in retinal research. Oxford: Pergamon, ally higher (65). A larger population study that includes 1990, pp.273–336. neuroimaging measures would be required to investi- 10. Coleston DM, Chronicle E, Ruddock KH, et al. Precor- gate the presence of brain lesions and the effect on tical dysfunction of spatial and temporal visual process- the PVER in people with migraine. ing in migraine. J Neurol Neurosurg Psychiatry 1994; 57: In summary, this study extends our previous know- 1208–1211. ledge of cortically evoked responses between migraine 11. McKendrick AM and Badcock DR. Contrast-processing events, by concurrently measuring retinal responses to dysfunction in both magnocellular and parvocellular both steady-state and transient stimulation. We con- pathways in migraineurs with or without aura. Invest Ophthalmol Vis Sci 2003; 44: 442–428. firm previous reports of abnormal cortically evoked 12. McKendrick AM and Sampson GP. Low spatial responses between migraine attacks. However, the frequency contrast sensitivity deficits in migraine are overlap between the migraine and control groups indi- not visual pathway selective. Cephalalgia 2009; 29: cates that these tests are unlikely to be of use as clinical 539–549. markers of the disease. PVER amplitude reduction was 13. Lewis RA, Vijayan N, Watson C, et al. Visual field loss in evident in both migraine subtypes at high temporal migraine. Ophthalmology 1989; 96: 321–326.
278 906 Cephalalgia 32(12)
14. Drummond PD and Anderson M. Visual field loss after 32. Marrelli A, Tozzi E, Porto C, et al. Spectral analysis of attacks of migraine with aura. Cephalalgia 1992; 12: visual potentials evoked by pattern-reversal checkerboard 349–352. in juvenile patients with headache. Headache 2001; 41: 15. De Natale RD, Polimeni D, Narbone MC, et al. Visual 792–797. field defects in migraine patients. In: Mills R (ed.) 33. Diener HC, Scholz E, Dichgans J, et al. Central effects of Perimetry update 1992/93: proceedings of the Xth drugs used in migraine prophylaxis evaluated by visual International Perimetric Society meeting. Amsterdam: evoked potentials. Ann Neurol 1989; 25: 125–130. Kugler Publications, 1993, pp.283–284. 34. Wollstein G, Garway-Heath DF and Hitchings RA. 16. McKendrick AM, Vingrys AJ, Badcock DR, et al. Visual Identification of early glaucoma cases with the scanning field losses in subjects with migraine headaches. Invest laser ophthalmoscope. Ophthalmology 1998; 105: Ophthalmol Vis Sci 2000; 41: 1239–1247. 1557–1563. 17. C¸ omog˘ lu S, Yarangu¨ meli A, Ko¨ z OG, et al. Glaucoma- 35. Headache Classification Subcommittee of the tous visual field defects in patients with migraine. International Headache Society. The international classi- J Neurol 2003; 250: 201–206. fication of headache disorders. 2nd edn. Cephalalgia 18. McKendrick AM and Badcock DR. Decreased visual 2004; 24(Suppl 1): 9–160. field sensitivity measured 1 day, then 1 week, after 36. Lipton RB, Stewart WF, Sawyer J, et al. Clinical utility migraine. Invest Ophthalmol Vis Sci 2004; 45: 106–170. of an instrument assessing migraine disability: the 19. McKendrick AM and Badcock DR. An analysis of the Migraine Disability Assessment (MIDAS) questionnaire. factors associated with visual field deficits measured with Headache 2001; 41: 854–861. flickering stimuli in-between migraine. Cephalalgia 2004; 37. Bach M and Speidel-Fiaux A. Pattern electroretinogram 24: 389–397. in glaucoma and ocular hypertension. Doc Ophthalmol 20. Flammer J, Pache M and Resink T. Vasospasm, its role 1989; 73: 173–181. in the pathogenesis of diseases with particular reference 38. Trick GL. Retinal potentials in patients with primary to the eye. Prog Retin Eye Res 2001; 20: 319–349. open-angle glaucoma: physiological evidence for tem- 21. Killer HE, Forrer A and Flammer J. Retinal vasospasm poral frequency tuning deficits. Invest Ophthalmol Vis during an attack of migraine. Retina 2003; 23: 253–254. Sci 1985; 26: 1750–1758. 22. Drance S, Anderson DR and Schulzer M. Risk factors 39. Odom JV, Bach M, Brigell M, et al. ISCEV standard for for progression of visual field abnormalities in normal- clinical visual evoked potentials. Doc Ophthalmol 2010; tension glaucoma. Am J Ophthalmol 2001; 131: 699–708. 120: 111–119. 23. Martinez A, Proupim N and Sanchez M. Retinal nerve 40. Holder GE, Brigell MG, Hawlina M, et al. ISCEV stand- fibre layer thickness measurements using optical coher- ard for clinical pattern electroretinography—2007 ence tomography in migraine patients. Br J Ophthalmol update. Doc Ophthalmol 2007; 114: 111–116. 2008; 92: 1069–1075. 41. Bach M and Hoffmann MB. Update on the pattern elec- 24. Principles and Practice of Clinical Electrophysiology of troretinogram in glaucoma. Optom Vis Sci 2008; 85: Vision. 2nd edn. Cambridge, MA: MIT Press, 2006. 386–395. 25. Shepherd AJ. Local and global motion after-effects are 42. Marcus DA and Soso MJ. Migraine and stripe-induced both enhanced in migraine, and the underlying mechan- visual discomfort. Arch Neurol 1989; 46: 1129–1132. isms differ across cortical areas. Brain 2006; 129: 43. Porciatti V and Ventura LM. Normative data for a user- 1833–1843. friendly paradigm for pattern electroretinogram record- 26. Karanovic O, Thabet M, Wilson HR, et al. Detection and ing. Ophthalmology 2004; 111: 161–168. discrimination of flicker contrast in migraine. Cephalalgia 44. Otto T and Bach M. Retest variability and diurnal effects 2011; 31: 723–736. in the pattern electroretinogram. Documenta Ophthalmol 27. Magis D, Ambrosini A, Bendtsen L, et al. Evaluation and 1996; 92: 311–323. proposal for optimalization of neurophysiological tests in 45. Kaufman D and Celesia GG. Simultaneous recording of migraine: part 1—electrophysiological tests. Cephalalgia pattern electroretinogram and visual evoked responses in 2007; 27: 1323–1338. neuro-ophthalmologic disorders. Neurology 1985; 35: 28. Ambrosini A, de Noordhout AM, Sandor PS, et al. Elec- 644–651. trophysiological studies in migraine: a comprehensive 46. Meigen T and Bach M. On the statistical significance review of their interest and limitations. Cephalalgia of electrophysiological steady-state responses. Doc 2003; 23(Suppl 1): 13–31. Ophthalmol 1999; 98: 207–232. 29. Shibata K, Osawa M and Iwata M. Simultaneous record- 47. Shibata K, Osawa M and Iwata M. Pattern reversal ing of pattern reversal electroretinograms and visual visual evoked potentials in migraine with aura and evoked potentials in migraine. Cephalalgia 1997; 17: migraine aura without headache. Cephalalgia 1998; 18: 742–747. 319–323. 30. Khalil NM. Investigations of visual function in migraine by 48. Coppola G, Parisi V, Fiermonte G, et al. Asymmetric visual evoked potentials and visual psychophysical tests. distribution of visual evoked potentials in patients with PhD Thesis, University of London, 1991. migraine with aura during the interictal phase. Eur J 31. Shibata K, Yamane K, Otuka K, et al. Abnormal visual Ophthalmol 2007; 17: 828–835. processing in migraine with aura: a study of steady-state 49. Logi F, Bonfiglio L, Orlandi G, et al. Asymmetric scalp visual evoked potentials. J Neurol Sci 2008; 271: 119–126. distribution of pattern visual evoked potentials during
279 Nguyen et al. 907
interictal phases in migraine. Acta Neurol Scand 2001; 58. Coppola G, Curra A, Sava SL, et al. Changes in 104: 301–307. visual-evoked potential habituation induced by hyper- 50. Tagliati M, Sabbadini M, Bernardi G, et al. Multichannel ventilation in migraine. J Headache Pain 2010; 11: visual evoked potentials in migraine. Electroencephalogr 497–503. Clin Neurophysiol 1995; 96: 1–5. 59. Coppola G, Cremers J, Gerard P, et al. Effects of light 51. Sand T, Zhitniy N, White LR, et al. Visual evoked poten- deprivation on visual evoked potentials in migraine with- tial latency, amplitude and habituation in migraine: a out aura. BMC Neurol 2011; 11: 91. longitudinal study. Clin Neurophysiol 2008; 119: 60. Hoeffken O, Stude P, Lenz M, et al. Visual paired- 1020–1027. pulse stimulation reveals enhanced visual cortex excit- 52. Boylu E, Domac FM, Kocer A, et al. Visual evoked ability in migraineurs. Eur J Neurosci 2009; 30: potential abnormalities in migraine patients. 714–720. Electromyogr Clin Neurophysiol 2010; 50: 303–308. 61. Coppola G, Vandenheede M, Di Clemente L, et al. 53. Khalil NM, Legg NJ and Anderson DJ. Long term Somatosensory evoked high-frequency oscillations decline of P100 amplitude in migraine with aura. reflecting thalamo-cortical activity are decreased in J Neurol Neurosurg Psychiatry 2000; 69: 507–511. migraine patients between attacks. Brain 2005; 128: 54. Kennard C, Gawel M, Rudolph Nde M, et al. Visual 98–103. evoked potentials in migraine subjects. Res Clin Stud 62. Coppola G, Pierelli F and Schoenen J. Habituation and Headache 1978; 6: 73–80. migraine. Neurobiol Learn Mem 2009; 92: 249–259. 55. Schoenen J, Wang W, Albert A, et al. Potentiation 63. Celesia GG, Kaufman D and Cone S. Effects of age and instead of habituation characterizes visual evoked poten- sex on pattern electroretinograms and visual evoked tials in migraine patients between attacks. Eur J Neurol potentials. Electroencephalogr Clin Neurophysiol 1987; 1995; 2: 115–122. 68: 161–171. 56. Sicuteri F. Headache as possible expression of deficiency 64. Bigal ME and Lipton RB. The epidemiology, burden, of brain 5-hydroxytryptamine (central denervation super- and comorbidities of migraine. Neurol Clin 2009; 27: sensitivity). Headache 1972; 12: 69–72. 321–334. 57. Afra J, Cecchini AP, De Pasqua V, et al. Visual evoked 65. Schurks M, Rist PM, Bigal ME, et al. Migraine and car- potentials during long periods of pattern-reversal stimu- diovascular disease: systematic review and meta-analysis. lation in migraine. Brain 1998; 121(Pt 2): 233–241. BMJ 2009; 339: b3914.
280 8 | Appendix B
This appendix includes supplementary material to Experiment 2 (Chapter 3).
8.1 Headache diary
A Headache Diary was provided to all migraine participants to assist in documenting the nature and timing of their attacks in between test sessions. The information was also used to monitor for any changes in migraine symptoms and medication usage.
281
282 8.2 Visual field pointwise analysis
8.2.1 Number of abnormal points across visual field
The following tables show the cumulative probabilities for determining the number of abnormal points to judge an entire visual field (Table 8.1), or portion of visual field (Tables 8.2 and 8.3), as abnormal (p < 0.05). These calculations were based on the 26 non-headache control participants (α = 0.04) in Experiment 2 (Chapter 3).
Table 8.1 – The cumulative probability (p) that a visual field consisted of at least n number of points below the lower€ limit for controls (α = 0.04), given the total number of test locations (N) across the entire visual field. Visual fields were considered abnormal overall (p < 0.05, denoted by bold text) if there were at least x number of statistically abnormal points.
p p p
n SAP (N = 101) TMP (N = 73) SWAP (N = 50)
0 0.016 0.051 0.130
1 0.068 0.154 0.271
2 0.142 0.232 0.276
3 0.195 0.229 0.184
4 0.199 0.167 0.090
5 0.161 0.096 0.035
6 0.107 0.045 0.011
7 0.061 0.018 0.003
8 0.030 0.006 0.001
9 0.013 0.002 < 0.001
x 8 6 5
283 Table 8.2 – The cumulative probability (p) that an upper quadrant consisted of at least n number of points below the lower limit for controls (α = 0.04), given the total number of test locations (N) in the quadrant. A quadrant was considered abnormal overall (p < 0.05, denoted by bold text) if there were at least x number of statistically abnormal points.
p p p
n SAP (N = 25) TMP (N = 17) SWAP (N = 12)
0 0.360 0.500 0.613
1 0.375 0.354 0.306
2 0.188 0.118 0.070
3 0.060 0.025 0.010
4 0.014 0.004 0.001
5 0.002 < 0.001 < 0.001
x 4 3 3
Table 8.3 – The cumulative probability (p) that a lower quadrant consisted of at least n number of points below the lower limit for controls (α = 0.04), given the total number of test locations (N) in the quadrant. A quadrant was considered abnormal overall (p < 0.05, denoted by bold text) if there were at least x number of statistically abnormal points.
p p p
n SAP (N = 24) TMP (N = 16) SWAP (N = 12)
0 0.375 0.520 0.613
1 0.375 0.347 0.306
2 0.180 0.108 0.070
3 0.055 0.021 0.010
4 0.012 0.003 0.001
5 0.002 < 0.001 < 0.001
x 4 3 3
284 8.2.2 Total number of depressed points
A ‘worst’ eye was chosen for analysis based on the total number of depressed points across SAP, TMP, and SWAP visual fields. The total number of depressed points for each control and migraine participant in Experiment 2 (Chapter 3) are shown below. The proportion of right eyes that were deemed the worst eye in the control (50%) and migraine (53%) were not different (chi-square test of proportions, p = 0.85).
Figure 8.1 – Total number of abnormal visual field locations, summed over SAP, TMP, and SWAP tests. Data from the right eyes are shown as white bars and data from the left eyes are shown as black bars.
8.3 Representative waveforms
Based on the relative change in amplitude from one visit to the other, the 25th, 50th, and 75th percentiles of change in PERG and PVER amplitude (raw waveforms and corresponding Fourier spectra, where applicable) are depicted in the following figures:
285
Figure 8.2 – Representative waveforms for Experiment 2, transient PERG. The 25th, 50th, and 75th percentiles of difference in P50 amplitude between first and second visits are shown for each group (control, MO, MA). The 25th percentile shows a reduction in amplitude, the 50th percentile shows almost no change in amplitude, and the 75th percentile shows an increase in amplitude at the second visit (closer to a migraine).
286
Figure 8.3 – Representative waveforms and corresponding Fourier spectra for Experiment 2, steady-state PERG. The 25th, 50th, and 75th percentiles of difference in 2F amplitude between first and second visits are shown for each group (control, MO, MA). The 25th percentile shows a reduction in amplitude, the 50th percentile shows almost no change in amplitude, and the 75th percentile shows an increase in amplitude at the second visit (closer to a migraine).
287
Figure 8.4 – Representative waveforms for Experiment 2, transient PVER. The 25th, 50th, and 75th percentiles of difference in P100 amplitude between first and second visits are shown for each group (control, MO, MA). The 25th percentile shows a reduction in amplitude, the 50th percentile shows almost no change in amplitude, and the 75th percentile shows an increase in amplitude at the second visit (closer to a migraine).
288
Figure 8.5 – Representative waveforms and corresponding Fourier spectra for Experiment 2, steady-state PVER. The 25th, 50th, and 75th percentiles of difference in 2F amplitude between first and second visits are shown for each group (control, MO, MA). The 25th percentile shows a reduction in amplitude, the 50th percentile shows almost no change in amplitude, and the 75th percentile shows an increase in amplitude at the second visit (closer to a migraine).
289 8.4 Extra statistical analyses
8.4.1 False negative rates
Table 8.4 shows that the false negative rates were the same between migraine and control groups at both visits (main effect of group: p > 0.05) and did not change with time of testing (interaction between group and visit: p > 0.05).
Table 8.4 – False negative (%) rates for visual field tests (median, minimum – maximum).
Controls Migraine
Test Visit 1 Visit 2 Visit 1 Visit 2 p-valuea p-valueb
SAP 0 (0 – 9) 0 (0 – 14) 0 (0 – 13) 0 (0 – 14) 0.07 0.37
TMP 0 (0 – 16) 5 (0 – 15) 5 (0 – 23) 5 (0 – 25) 0.12 0.78
SWAP 0 (0 – 17) 0 (0 – 0) 0 (0 – 14) 0 (0 – 14) 0.08 0.85 a p-values represent the RM-ANOVA main effect of group (control, migraine) across both visits b p-values represent the RM-ANOVA interaction between group (control, migraine) and visit (visit 1, visit 2)
8.4.2 Exclusion of pre-attack data
Tables 8.5 to 8.7 demonstrate that the electrophysiological results were not changed when data were excluded from the single individual who experienced a migraine one day after testing. Similarly, visual field performance was unaffected by exclusion of pre-attack data, based on analysis of the global indices (Tables 8.8 and 8.9).
290 Table 8.5 – Summary of PERG parameters (mean ± standard deviation), excluding data from the single participant who was tested one day before a migraine.
Controls Migraine
Parameter Visit 1 Visit 2 Visit 1 Visit 2 p-valuea
Transient P50 amplitude (µV) 6.2 ± 1.4 6.6 ± 1.5 6.5 ± 1.4 6.4 ± 1.7 0.30
N95 amplitude (µV) 10.2 ± 2.6 9.8 ± 2.1 11.5 ± 3.3 9.9 ± 2.1 0.16
N35 time (ms) 22 ± 3 22 ± 3 21 ± 2 22 ± 3 0.43
P50 time (ms) 51 ± 3 51 ± 2 51 ± 2 51 ± 2 0.74
N95 time (ms) 104 ± 5 106 ± 4 103 ± 4 106 ± 4 0.96
Steady-state 2F amplitude (µV) 3.2 ± 0.5 3.3 ± 0.6 3.1 ± 1.0 3.2 ± 0.6 0.99
2F phase (π rads) 2.1 ± 0.1 2.1 ± 0.1 2.1 ± 0.1 2.1 ± 0.1 0.38
PERG ratio 1.2 ± 0.2 1.2 ± 0.2 1.3 ± 0.3 1.3 ± 0.3 0.64 a p-values represent the RM-ANOVA interaction between group (control, migraine) and visit (visit 1, visit 2)
Table 8.6 – Summary of PVER parameters (mean ± standard deviation), excluding data from the single participant who was tested one day before a migraine.
Controls Migraine
Parameter Visit 1 Visit 2 Visit 1 Visit 2 p-valuea
Transient P100 amplitude (µV) 11.3 ± 5.0 10.6 ± 5.3 12.1 ± 5.4 13.0 ± 5.6 0.03 †
N135 amplitude (µV) 10.0 ± 2.5 9.8 ± 2.1 16.5 ± 6.6 17.0 ± 7.3 0.13
N75 time (ms) 72 ± 7 71 ± 8 71 ± 4 69 ± 7 0.61
P100 time (ms) 103 ± 6 102 ± 6 102 ± 3 102 ± 4 0.31
N135 time (ms) 146 ± 15 144 ± 15 143 ± 8 146 ± 8 0.07
Steady-state 2F amplitude (µV) 3.9 ± 1.9 3.7 ± 2.0 2.1 ± 1.4 ** 1.7 ± 1.0 ** 0.40
2F phase (π rads) 1.5 ± 0.4 1.5 ± 0.3 1.5 ± 0.2 1.6 ± 0.4 0.85 a p-values represent the RM-ANOVA interaction between group (control, migraine) and visit (visit 1, visit 2)
† paired t-tests: controls t25 = 1.63, p = 0.12; migraine t15 = 1.47, p = 0.16 ** denotes significant reduction in steady-state PVER amplitude in migraine group at both visits [RM-ANOVA main effect of group: F (1, 39) = 12.13, p = 0.001]
291 Table 8.7 – PVER interhemispheric asymmetry (median, minimum – maximum), excluding data from the single participant who was tested one day before a migraine.
Controls Migraine
Amplitude a a Visit 1 Visit 2 p-value Visit 1 Visit 2 p-value asymmetry (%)
Transient P100 18 (1 – 42) 21 (3 – 62) 0.24 16 (2 – 56) 14 (2 – 43) 0.31
Transient N135 22 (3 – 49) 12 (2 – 37) 0.10 14 (2- 70) 15 (2 – 37) 0.74
Steady-state 2F 15 (2 – 41) 20 (3 – 51) 0.29 18 (3 – 60) 26 (3 – 75) 0.47 a p-values represent the effect of visit (paired Wilcoxon signed rank tests)
Table 8.8 – Average/Mean Defect (mean ± standard deviation), excluding data from the single participant who was tested one day before a migraine.
Controls Migraine
Test Global index Visit 1 Visit 2 Visit 1 Visit 2 p-valuea
SAP Average Defect (dB) -0.9 ± 1.1 0 ± 1.3 -0.8 ± 1.0 -0.6 ± 1.2 0.03
TMP Average Defect (dB) -3.2 ± 1.0 -3.2 ± 1.1 -3.4 ± 1.3 -3.6 ± 1.2 0.69
SWAP Mean Defect (dB) 3.1 ± 2.9 2.9 ± 3.0 3.3 ± 2.9 4.1 ± 2.8 0.10 a p-values represent the RM-ANOVA interaction between group (control, migraine) and visit (visit 1, visit 2)
Table 8.9 – Pattern Defect and Loss Variance (median, minimum – maximum), excluding data from the single participant who was tested one day before a migraine.
Controls Migraine
Test Visit 1 Visit 2 p-valuea Visit 1 Visit 2 p-valuea
1.4 1.7 3.2 2.5 SAP 0.16 0.92 (0 – 4.6) (0 – 5.5) (0 – 20.4) (0 – 20.3)
2.3 1.6 2.1 3.5 TMP 0.12 0.18 (0 – 8.1) (0 – 9.6) (0 – 18.3) (0 – 16.3)
7.8 6.4 14.0 12.7 SWAP 0.36 0.24 (2.2 – 33.8) (2.0 – 20.5) (3.5 – 114.0) (3.7 – 116.0) a p-values represent the effect of visit (paired Wilcoxon signed rank tests)
292 8.4.3 Correlations with migraine characteristics
Tables 8.10 to 8.13 indicate that there was no relationship between self-reported migraine characteristics and electrophysiology and visual field parameters, averaged across both test visits.
Table 8.10 – Relationship between migraine characteristics and SAP.
Pattern Defect Depressed points
R p-valuea R p-valuea
Age at first migraine (years) 0.09 0.75 -0.04 0.89
Years of migraine 0.40 0.11 0.22 0.39
Weeks between migraines 0.07 0.79 0.18 0.49
Migraines in past year -0.09 0.74 -0.27 0.29
Estimated number of lifetime attacks 0.09 0.72 -0.15 0.55
MIDAS questionnaire score (days) -0.26 0.32 -0.47 0.06
Headache duration (hours) 0.05 0.85 -0.03 0.90 a p-values represent the significance of Spearman rank correlations
Table 8.11 – Relationship between migraine characteristics and TMP.
Pattern Defect Depressed points
R p-valuea R p-valuea
Age at first migraine (years) -0.05 0.84 -0.12 0.63
Years of migraine 0.16 0.54 0.19 0.46
Weeks between migraines -0.27 0.29 -0.28 0.28
Migraines in past year 0.20 0.43 0.21 0.43
Estimated number of lifetime attacks 0.27 0.29 0.29 0.26
MIDAS questionnaire score (days) 0.30 0.25 0.16 0.53
Headache duration (hours) 0.19 0.46 -0.04 0.89 a p-values represent the significance of Spearman rank correlations
293 Table 8.12 – Relationship between migraine characteristics and SWAP.
Pattern Defect Depressed points
R p-valuea R p-valuea
Age at first migraine (years) 0.24 0.35 0.28 0.28
Years of migraine 0.28 0.28 0.14 0.60
Weeks between migraines -0.07 0.78 -0.05 0.84
Migraines in past year 0.09 0.73 0.04 0.87
Estimated number of lifetime attacks 0.19 0.47 0.07 0.80
MIDAS questionnaire score (days) -0.09 0.72 -0.16 0.54
Headache duration (hours) 0.15 0.55 -0.10 0.70 a p-values represent the significance of Spearman rank correlations
Table 8.13 – Relationship between migraine characteristics and steady-state PVER amplitude.
R p-valuea
Age at first migraine (years) 0.09 0.73
Years of migraine -0.15 0.55
Weeks between migraines -0.15 0.57
Migraines in past year 0.13 0.63
Estimated number of lifetime attacks 0.09 0.73
MIDAS questionnaire score (days) 0.54 0.02
Headache duration (hours) 0.11 0.67 a p-values represent the significance of Spearman rank correlations
8.5 Published work
The following journal article has been published from the work arising from Experiment 2 (Chapter 3).
294 XML Template (2013) [20.8.2013–3:23pm] [1–16] //blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/CEPJ/Vol00000/130154/APPFile/SG-CEPJ130154.3d (CEP) [PREPRINTER stage]
Original Article
Cephalalgia 0(0) 1–16 ! International Headache Society 2013 The effect of duration post-migraine on Reprints and permissions: sagepub.co.uk/journalsPermissions.nav visual electrophysiology and visual field DOI: 10.1177/0333102413498939 performance in people with migraine cep.sagepub.com
Bao N Nguyen, Algis J Vingrys and Allison M McKendrick
Abstract Purpose: In between migraine attacks, some people show visual field defects that are worse when measured closer to the end of a migraine event. In this cohort study, we consider whether electrophysiological responses correlate with visual field performance at different times post-migraine, and explore evidence for cortical versus retinal origin. Methods: Twenty-six non-headache controls and 17 people with migraine performed three types of perimetry (static, flicker and blue-on-yellow) to assess different aspects of visual function at two visits conducted at different durations post-migraine. On the same days, the pattern electroretinogram (PERG) and visual evoked response (PVER) were recorded. Results: Migraine participants showed persistent, interictal, localised visual field loss, with greater deficits at the visit nearer to migraine offset. Spatial patterns of visual field defect consistent with retinal and cortical dysfunction were identified. The PERG was normal, whereas the PVER abnormality found did not change with time post-migraine and did not correlate with abnormal visual field performance. Conclusions: Dysfunction on clinical tests of vision is common in between migraine attacks; however, the nature of the defect varies between individuals and can change with time. People with migraine show markers of both retinal and/or cortical dysfunction. Abnormal visual field sensitivity does not predict abnormality on electrophysiological testing.
Keywords Migraine, contrast, visual fields, visual evoked potential, electroretinogram
Date received: 2 April 2013; revised: 31 May 2013; accepted: 15 June 2013
Introduction involvement of the pre-cortical visual pathways. Case Migraine is a common neurological disorder involving studies demonstrate retinal vascular involvement in vision. Many studies have identified abnormal visual some individuals (e.g. (28)), and reduced retinal nerve function in between migraine attacks (the interictal fibre layer thickness (29) and transient retinal vaso- period). These include perceptual measures of cortical spasm (30) have been associated with migraine. visual processing (e.g. (1–4)), as well as electrophysi- Several studies report performance differences on psy- ology (e.g. (5–14)) and visual field assessment using chophysical tasks that assess pre-cortical vision (4,31– static (15–22), flickering (20,23,24), and blue-on- 35). Furthermore, the spatial pattern of visual field yellow perimetry (25,26). defects can resemble retinal (e.g. monocular and arcu- Previous literature does not suggest a single, ate (15,18,20,21,23,25,26)) or cortical (e.g. bilateral and common anatomical locus for visual anomalies in migraine. Brain neuroimaging has demonstrated struc- tural changes in both primary visual cortex (V1) and Department of Optometry and Vision Sciences, The University of extrastriate areas (for a review, see Schwedt and Melbourne, Australia Dodick (27)). Electrophysiology suggests cortical Corresponding author: involvement, as abnormal cortical evoked potentials Allison M McKendrick, Department of Optometry and Vision Sciences, occur concurrently with normal retinal responses The University of Melbourne, Parkville, Victoria, 3010, Australia. (6,8,14). However, there is also evidence for Email: [email protected]
295 XML Template (2013) [20.8.2013–3:23pm] [1–16] //blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/CEPJ/Vol00000/130154/APPFile/SG-CEPJ130154.3d (CEP) [PREPRINTER stage]
2 Cephalalgia 0(0)
homonymous (17,19,22)) dysfunction in different according to the tenets of the Declaration of Helsinki, people. These interictal visual field defects do not was obtained prior to participation. only occur in people who experience visual aura The study included people with migraine and non- during their migraine attacks. headache controls. Participants were recruited from 75 A challenge for experiments considering the anatom- participants in a previous cross-sectional study (14), ical locus of visual dysfunction in migraine is the fact who were all asked to return for a second test. After that migraine is an episodic condition. Visual function regular follow-up attempts were made by phone and can vary with time – both in the lead up to a migraine email from June 2010 to July 2012, 17 people with (10,36,37) and post-migraine (9,16,19,20,24). The migraine (11 MO, 6 MA) and 26 non-headache controls increase (36,37) and normalisation (10) of cortical returned. All participants were screened to satisfy the evoked potentials in the pre-attack period are presumed following inclusion criteria: best corrected visual acuity to reflect physiological changes involved in the build up 6/7.5 (logMAR), subjective refraction within 5.00D to a migraine event, such as the normalisation of cor- sphere! and 2.00D astigmatism, intraocular pressureÆ tical excitability (38,39) or the increase in serotonin <21 mmHgÀ by Goldmann applanation tonometry, immediately before an attack (37,40). In contrast, age-normal findings on slitlamp biomicroscopy, oph- visual field defects are worse the day after a migraine thalmoscopy and optic nerve head imaging with the (24) and gradually improve over time (16,19,20), which Heidelberg Retinal Tomograph (HRT), and no sys- suggests that they may be sequelae of migraine. temic disease or medications known to affect visual In this study, we compare visual fields and electro- function or neurological state, including prophylactive physiology in the same individuals, measured on the migraine medications. The control (19–46 years) and same day, at different time-points after migraine. To migraine (19–43 years) groups did not differ in age our knowledge, this is the first study to directly com- (Mann Whitney rank sum test, p 0.12). Neither was pare visual field assessment with electrophysiology in there a group difference in¼ global rim area the same migraine cohort. We consider the anatomical (F(2,71) 1.08, p 0.34) or volume (F(2,71) 0.98, locus of abnormalities, as inferred from the spatial p 0.38)¼ of the optic¼ nerve head, which are two¼ HRT pattern and binocularity of visual field defects, and parameters¼ that correlate with perimetric indices from comparison of simultaneously recorded pattern describing generalised and localised visual field loss in electroretinogram (PERG) and visual evoked response people with glaucoma (41). (PVER). Participants completed a clinical interview, headache questionnaire and the Migraine Disability Assessment Score (MIDAS) questionnaire (42) to describe their Methods migraine attacks, where applicable (Table 1). The Participants MIDAS questionnaire score measures the number of days in the preceding three months where migraine The study was approved by the Human Research attacks resulted in reduced productivity in tasks of Ethics Committee of the University of Melbourne daily living. Scores are interpreted as minimal (Grade (HREC #0932638). Written informed consent, 1, score 0–5), mild (Grade 2, score 6–10), moderate
Table 1. Summary of self-reported migraine characteristics (median, range).
Migraine characteristics
MO MA Statistic p
Days since last migraine at Visit 1 18 (7–150) 25 (7–200) U 26.0 0.51 ¼ Days since last migraine at Visit 2 3 (1–5) 2 (1–6) U 24.0 0.38 ¼ Age at first migraine (years) 15 (4–17) 12 (10–30) U 21.0 0.24 ¼ Years of migraine 13 (3–23) 20 (7–30) t 0.77 0.45 15 ¼ Migraine attacks in past year 8 (1–50) 5 (1–50) U 17.5 0.39 ¼ Weeks between migraine attacks 3 (1–20) 6 (1–24) U 22.0 0.29 ¼ Estimated number of lifetime attacks 100 (30–550) 89 (14–1300) U 25.0 0.45 ¼ MIDAS questionnaire score (days) 20 (0–49) 3 (1–4) U 7.0 0.010 ¼ Headache duration (hours) 12 (2–72) 8 (2–48) U 26.0 0.51 ¼ Independent sample t-tests and Mann Whitney rank sum tests comparing the migraine characteristics between groups are provided. MIDAS: Migraine Disability Assessment Score; MO: migraine without aura; MA: migraine with aura.
296 XML Template (2013) [20.8.2013–3:23pm] [1–16] //blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/CEPJ/Vol00000/130154/APPFile/SG-CEPJ130154.3d (CEP) [PREPRINTER stage]
Nguyen et al. 3
(Grade 3, score 11–20) or severe disability (Grade 4, 199 days (Figure 1; median 16 days). Control partici- score 21 ). Migraine participants reported symptoms pants completed two sessions at least one day apart (headache,þ nausea/vomiting, photophobia, phonopho- (median 18 days, range 1–132 days). bia) that fulfilled the International Headache Society Increased PVER amplitude has been reported in the criteria (43) for migraine without aura (MO) and pre-attack period, up to 72 hours before a migraine (37). migraine with aura (MA). The MO and MA groups Prodromal symptoms, including fatigue and difficulty were pooled, as both groups demonstrate similar concentrating, commonly occur up to 48 hours before visual field losses (16,18,20,21,23–25). Control partici- an attack (44) and may affect a person’s visual field per- pants had never had a migraine and were free from formance. To address this, all participants were con- regular headaches (less than four in the past year). tacted (by phone or email) after each test session. This follow-up found that the majority of participants did not Timing of the test visits have a migraine within 72 hours of each test session. One of the 17 participants experienced a migraine the day Each session lasted up to three hours. For people with after the first test session. Data from this participant migraine, the first visit was scheduled at least one week have been represented as cross symbols in Figures 1–3. post-migraine. The second visit was scheduled as close Excluding the data from this individual from statistical as practicable, but at least one day, after the cessation analyses did not change our conclusions. of migraine symptoms (maximum six days post- migraine). The difference in the number of days post- Visual field tests migraine between the two visits ranged from three to Visual field tests were always conducted first because of possible ocular discomfort following electrode place- ment for PERG recordings. Three different visual field tests were included. Standard automated perimetry (SAP) is the standard perimetric technique and is most commonly encountered in clinical practice. Participants completed SAP first, as it is well tolerated and generally 100 easiest for a naı¨ ve observer to learn. Temporal modu- lation perimetry (TMP) and short-wavelength auto- mated perimetry (SWAP) were conducted next, in random order, as visual field defects in people with migraine have been identified using TMP (20,24) and SWAP (25) that are not measurable on SAP. These different forms of perimetry test different aspects of 10 visual processing, with flicker perimetry preferentially assessing magnocellular pathways (45) and SWAP assessing the blue-on-yellow (or koniocellular) system
Days since last migraine (46). SAP is non-visual pathway selective (47). SAP and TMP were performed on the Medmont M- 700 perimeter (Medmont Pty Ltd., Camberwell, Victoria, Australia), which has been described else- where (48). In brief, the stimuli (!max 565 nm, max 2 ¼ 1 luminance 320 cd/m ) are 0.43 (Goldmann size III) light-emitting diodes presented on a background lumi- nance of 3.2 cd/m2 (CIE 1931 x 0.53, y 0.42) and ¼ ¼ Visit 1 Visit 2 arranged in concentric rings. SAP thresholds were mea- sured using the Central Threshold test at 103 locations at 1,3,6, 10, 15, 22 and 30 eccentricities. For Figure 1. Days since last migraine at the two test visits for the TMP, the Auto-Flicker test was conducted at 73 loca- migraine participants. MO participants are shown as filled square tions at 1 ,3,6, 10 , 15 and 22 . This test varies the symbols, whereas MA participants are shown as filled diamond symbols. The MO participant who was tested one day before a temporal frequency of the flickering stimuli with retinal migraine is shown as a cross symbol. Visit 1 was scheduled at eccentricity (18 Hz, 1–3; 16 Hz, 6; 12 Hz, 10–15; least seven days after a migraine. Visit 2 was scheduled at a time 9 Hz, 22). Stimuli were presented for 200 ms (SAP) closer after a migraine (within six days). and 800 ms (TMP) durations. SWAP was performed MO: migraine without aura; MA: migraine with aura. on the Octopus 101 perimeter (Haag-Streit Inc.,
297 XML Template (2013) [20.8.2013–3:23pm] [1–16] //blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/CEPJ/Vol00000/130154/APPFile/SG-CEPJ130154.3d (CEP) [PREPRINTER stage]
4 Cephalalgia 0(0)
Controls Migraine (a)
9
6 V) µ
3
0
(b) 30
25
20 V) Amplitude ( µ
15 Amplitude ( 10
5
0
Visit 1 Visit 2 Visit 1 Visit 2
Figure 2. PVER amplitudes at the two test visits. (a) Steady-state PVER. (b) Transient PVER. Individual data are presented for the control (unfilled symbols) and migraine (MO: filled squares, MA: filled diamonds) participants. The MO participant who was tested one day before a migraine is shown as a cross symbol. Visit 1 was scheduled at least seven days after a migraine. Visit 2 was scheduled at a time closer after a migraine (within six days). Error bars represent the group mean 95% CI of the mean. PVER: pattern visual evoked response; MO: migraine without aura; MA: migraine withÆ aura.
Koeniz, Switzerland), a detailed description of which Dynamic strategy (50) at 52 locations at 3,9, 15 has been given previously (49). Blue (! 440 nm) and 21 eccentricity. max ¼ test stimuli of 1.72 (Goldmann size V) were projected Participants had a brief practice before testing. Tests for 200 ms against a yellow background (100 cd/m2), to with false-positive or false-negative rates above 30% which participants adapted for at least three minutes were excluded. The automated blind-spot monitor iden- before testing. Thresholds were measured using the tified fixation losses exceeding 30% in four control and
298 XML Template (2013) [20.8.2013–3:23pm] [1–16] //blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/CEPJ/Vol00000/130154/APPFile/SG-CEPJ130154.3d (CEP) [PREPRINTER stage]
Nguyen et al. 5
Controls Migraine (a) 24
18
12
Pattern Defect (dB) 6
0
(b) 24
18
12
Pattern Defect (dB) 6
0
(c)
100 ) 2
10 Loss Variance (dB
1 Visit 1 Visit 2 Visit 1 Visit 2
Figure 3. Global indices of localised visual field loss at the two test visits. (a) SAP Pattern Defect. (b) TMP Pattern Defect. (c) SWAP Loss Variance. Individual data are presented for the control (unfilled symbols) and migraine (MO: filled squares, MA: filled diamonds) participants. The MO participant who was tested one day before a migraine is shown as a cross symbol. Visit 1 was scheduled at least seven days after a migraine. Visit 2 was scheduled at a time closer after a migraine (within six days). Horizontal lines represent the group median. MO: migraine without aura; MA: migraine with aura; SAP: standard automated perimetry; TMP: temporal modulation perimetry; SWAP: short-wavelength automated perimetry.
299 XML Template (2013) [20.8.2013–3:23pm] [1–16] //blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/CEPJ/Vol00000/130154/APPFile/SG-CEPJ130154.3d (CEP) [PREPRINTER stage]
6 Cephalalgia 0(0)
four migraine participants. However, continuous moni- integrity of the retinocortical pathway (53) and toring of the limbal position by direct visual inspection other brain areas (54). The PERG and PVER were (Medmont) or via video camera (Octopus) confirmed recorded simultaneously to rule out cortical dysfunc- steady fixation. tion arising from the presence of a retinal abnormal- ity. Both ‘transient’ (<3 Hz stimulation) and ‘steady- Visual field analysis state’ responses to higher temporal frequency stimulation ( 4 Hz) were included, as visual field sen- The global indices generated by the perimeter were ana- sitivity is worse! the day after a migraine when tested lysed. The Medmont perimeter returns Average Defect with flickering stimuli (TMP) (24). The steady-state and Pattern Defect, whereas the Octopus perimeter response is presumed to share similar neural sub- returns Mean Defect and Loss Variance. These indices strates to behavioural measures of temporal process- are determined relative to a proprietary age-matched ing (flicker) (55). normative database and describe generalised and loca- The protocol for simultaneous PERG and PVER lised visual field loss respectively. has been described in detail elsewhere (14). Responses Global indices provide single summary statistics for were recorded monocularly according to ISCEV visual field performance but do not illustrate which loca- standards (52,53) using the Espion (Diagnosys LLC, tions are abnormal across the visual field. To establish a Cambridge, UK). Electrode impedance was generally point-wise assessment of visual field abnormality, we below 5 kOhms and did not exceed 10 kOhms. determined two-sided empirical confidence limits of sen- Participants fixated on a 0.5 diameter red square sitivity at each visual field location (20), based on our 26 in the centre of the screen (Sony G520 21-inch control participants. We used our controls because CRT monitor: frame rate 100 Hz, resolution people with migraine are not excluded from the propri- 1024 786 pixels) positioned 50 cm away. The stimu- Â etary databases. Locations at and immediately above lus was a square-wave checkerboard (31 square field, 2 and below the blindspot were excluded. As visual field 52 cd/m mean luminance, 96% contrast, 0.8 checks), outcomes are non-parametric (51), locations where sen- counter-phased at 1 Hz (2 reversals/second, 250 ms sitivity was lower than the eighth percentile limit (second epoch, ‘transient’ response) or 8.3 Hz (16.7 reversals/ worst-performing control) were considered ‘depressed’, second, 480 ms epoch, ‘steady-state’ response). Stimuli and vice versa for ‘better’ points (upper 92nd percentile were presented using an interleaved block design to limit). In the same way, point-wise confidence limits balance the effect of fatigue on recordings. were determined for the change in sensitivity between Two-hundred signals were amplified, bandpass-fil- the first and second visits, where a negative change indi- tered (1.25–100 Hz), and digitised (1000 Hz) to 16-bit cated a reduction in sensitivity at the second visit. resolution. Timing and amplitude measures were Assuming that thresholds at individual locations are extracted. In compliance with ISCEV standards independent, visual fields were judged to be abnormally (52,53), peak times were measured for the positive depressed (p < 0.05) if there were at least eight locations components of the PERG (P50) and PVER (P100). below our control group lower limit (p < 0.04 for a single Peak-to-peak amplitudes were measured for the two location) out of a total 101 test points on SAP, 6 of 73 on neural signals closest in succession along the visual TMP and 5 of 50 on SWAP (see Appendix, Table A1). pathway, representing activity of the retinal ganglion The fellow eye was also examined to classify whether cells (PERG P50-N95 amplitude (56)) and V1 the pattern of defect was homonymous. Two (PVER N75-P100 amplitude (54)). The different approaches were used: (1) visual inspection, to see if components were identifiable on all transient wave- locations of depressed sensitivity were in the same forms collected. Similarly, the amplitude and phase hemifield in both eyes and respected the vertical mid- at the second harmonic (16.7 Hz) of the steady-state line; and (2) quadrant analysis (1), where a quadrant PERG and PVER, reflecting retinal ganglion cell was classified as abnormal (p < 0.05) if there were at (56) and primary visual cortical activity (57), respect- least four SAP, three TMP or three SWAP locations ively, were determined by Discrete Fourier that were depressed within that quadrant (see Transform. Decreased phase values correspond to Appendix, Table A1). When the same quadrant was signal delays in the time domain. Steady-state classified as abnormal in both eyes, using either criter- responses below noise levels at neighbouring frequen- ion, the deficit was considered homonymous. cies (14.6 and 18.8 Hz) (58) were removed from the dataset. PVER interhemispheric asymmetry Pattern electrophysiology (7,9,11,13), which may be related to the laterality of the migraine headache (11) or aura (7,9), was The PERG reflects retinal ganglion cell activity (52), defined as the percentage difference in amplitude whereas the PVER measures V1 function and between the right and left hemispheres.
300 XML Template (2013) [20.8.2013–3:23pm] [1–16] //blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/CEPJ/Vol00000/130154/APPFile/SG-CEPJ130154.3d (CEP) [PREPRINTER stage]
Nguyen et al. 7
Statistical analysis level was adjusted using a Holm-Bonferroni correction for multiple comparisons (59). For control and migraine participants, a worst eye was chosen for analysis based on the total number of abnor- mal points across all visual fields. Statistical compari- Results sons were performed using SPSS version 20.0 (SPSS Changes in electrophysiology with time post- Inc., Chicago, IL, United States). Data were tested to migraine confirm statistical normality (Shapiro-Wilk normality test) and homogeneity of variances (Mauchly’s test of We find differences in PVER amplitude between sphericity). Repeated-measures analyses of variance migraine and control groups depending on the compo- considered group differences (RM-ANOVA, a 0.05) nent analysed (Table 2; group component interaction: nested within visit (visit 1, 2) and test (transient, steady-¼ F(1,40) 7.92, p 0.008). Separate component analyses state) or perimeter (SAP, TMP, SWAP). Where the indicated¼ reduced¼ steady-state PVER (Figure 2(a); assumption of sphericity was violated, the degrees of F(1,40) 11.4, p 0.002), but normal transient PVER freedom were amended using a Huynh-Feldt correc- amplitudes¼ (Figure¼ 2(b); F(1,41) 1.37, p 0.25) in the tion. Paired t-tests, or Wilcoxon signed rank tests migraine group. Our data further¼ demonstrate¼ that where the data were non-Gaussian, were used to test PVER amplitude did not change at the second visit; for within-individual changes between visits. The alpha that is, closer to a migraine. Comparisons between the
Table 2. Summary of retinal (PERG) and cortical (PVER) electrophysiological measures (mean standard deviation) at the two test visits. Æ
Control Migraine
Visit 1 Visit 2 Visit 1 Visit 2 RM-ANOVA group comparisons
PERG amplitude (kV) Transient response 10.2 2.63 9.79 2.05 11.6 3.22 10.1 2.21 Group: F(1,41) 1.01, p 0.32 Æ Æ Æ Æ ¼ ¼ Steady-state 3.23 0.52 3.28 0.62 3.19 0.96 3.20 0.61 Group component: F(1,41) 2.76, p 0.10 Â ¼ ¼ response Æ Æ Æ Æ Group visit: F(1,41) 1.36, p 0.25 Â ¼ ¼ PERG timing Transient peak 51 3 51 2 51 2 51 2 Group: F(1,41) 0.65, p 0.43 time (ms) Æ Æ Æ Æ Group component:¼ F¼(1,41) 0.39, p 0.54 Â ¼ ¼ Steady-state 6.03 0.22 5.81 0.42 6.01 0.24 5.91 0.30 Group visit: F(1,41) 0.21, p 0.65 Â ¼ ¼ phase (rads) Æ Æ Æ Æ PVER amplitude (kV) Transient response 11.3 5.04 10.6 5.30 12.9 4.80 13.8 5.11 Group: F(1,40) 0.15, p 0.70 Æ Æ Æ Æ Group component:¼ F¼(1,40) 7.92, p 0.008**Â ¼ Group¼ visit: F(1,40) 4.04, p 0.051 Â ¼ ¼ Steady-state response 3.89 1.88 3.72 1.95 2.22 1.37 1.87 0.97 Æ Æ Æ Æ PVER timing Transient peak 103 6 102 6 103 6 102 4 Group: F(1,40) 0.03, p 0.87 time (ms) Æ Æ Æ Æ Group component:¼ F¼(1,40) 0.02, p 0.90 Â ¼ ¼ Steady-state 7.83 1.11 7.94 1.01 7.78 0.48 7.57 1.19 Group visit: F(1,40) 0.95, p 0.34 Â ¼ ¼ phase (rads) Æ Æ Æ Æ PVER/PERG ratio Transient 1.19 0.52 1.12 0.52 1.24 0.60 1.23 0.53 Group: F(1,40) 1.95, p 0.17 Æ Æ Æ Æ Group component:¼ F¼(1,40) 11.7, p 0.001**Â ¼ Group¼ visit: F(1,40) 0.05, p 0.82 Â ¼ ¼ Steady-state 1.25 0.63 1.19 0.67 0.74 0.51 0.60 0.48 Æ Æ Æ Æ Visit 1 was scheduled at least seven days after a migraine. Visit 2 was scheduled at a time closer after a migraine (within six days). RM-ANOVAs comparing the electrophysiological measures between groups are provided. **Denotes significance using Holm-Bonferroni correction for multiple comparisons, starting at p < 0.01. PERG: pattern electroretinogram; PVER: pattern visual evoked response; RM-ANOVA: repeated-measures analysis of variance.
301 XML Template (2013) [20.8.2013–3:23pm] [1–16] //blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/CEPJ/Vol00000/130154/APPFile/SG-CEPJ130154.3d (CEP) [PREPRINTER stage]
8 Cephalalgia 0(0)
Table 3. Summary of PVER amplitude interhemispheric asymmetry (median, range) at the two test visits.
Control Migraine
Visit 1 Visit 2 Paired tests Visit 1 Visit 2 Paired tests
Transient PVER asymmetry (%) 17 (1–37) 21 (3–62) Z 1.18, p 0.24 16 (2–39) 16 (2–43) Z 0.40, p 0.69 ¼À ¼ ¼À ¼ Steady-state PVER asymmetry (%) 15 (2–41) 20 (3–51) Z 1.06, p 0.29 22 (3–60) 26 (3–75) Z 0.87, p 0.38 ¼À ¼ ¼À ¼ Visit 1 was scheduled at least seven days after a migraine. Visit 2 was scheduled at a time closer after a migraine (within six days). Paired Wilcoxon signed rank tests comparing the asymmetry measures between visits are provided. PVER: pattern visual evoked response.
Table 4. Summary of visual field global indices at the two test visits. (a) Average/Mean Defect (mean standard deviation) and (b) Pattern Defect/Loss Variance (median, range). Æ
Control Migraine
Visit 1 Visit 2 Paired tests Visit 1 Visit 2 Paired tests
(a) Average/Mean Defect SAP Average 0.85 1.14 0.04 1.26 t25 4.80, 0.87 1.02 0.66 1.13 t16 1.08, Defect (dB) À Æ À Æ p¼< 0.001** À Æ À Æ p¼ 0.30 ¼ TMP Average 3.24 1.04 3.20 1.09 t25 0.27, 3.40 1.23 3.56 1.17 t25 0.65, Defect (dB) À Æ À Æ p¼ 0.79 À Æ À Æ p¼ 0.52 ¼ ¼ SWAP Mean 3.08 2.90 2.92 2.98 t25 0.77, 3.31 2.85 3.94 2.71 t25 1.39, Defect (dB) Æ Æ p¼ 0.45 Æ Æ p¼ 0.18 ¼ ¼ (b) Pattern Defect/Loss Variance SAP Pattern 1.44 (0–4.58) 1.67 (0–5.49) Z 1.41, 3.14 (0–20.4) 2.02 (0–20.3) Z 1.40, Defect (dB) ¼Àp 0.16 ¼Àp 0.16 ¼ ¼ TMP Pattern 2.30 (0–8.14) 1.57 (0–9.62) Z 1.56, 2.25 (0–18.3) 4.26 (0–16.3) Z 0.05, Defect (dB) ¼Àp 0.12 ¼Àp 0.96 ¼ ¼ SWAP Loss 7.75 (2.20–33.8) 6.35 (2.00–20.5) Z 0.92, 13.6 (3.50–113.9) 11.2 (3.70–115.5) Z 0.26, Variance (dB2) ¼Àp 0.36 ¼Àp 0.80 ¼ ¼ Visit 1 was scheduled at least seven days after a migraine. Visit 2 was scheduled at a time closer after a migraine (within six days). Paired t-tests and Wilcoxon signed rank tests comparing the global indices between visits are provided. **Denotes significance using Holm-Bonferroni correction for multiple comparisons, p < 0.004. SAP: standard automated perimetry; TMP: temporal modulation perimetry; SWAP: short-wavelength automated perimetry.
two visits were performed using paired t-tests and none Visual field changes with time post-migraine was found to be significant (steady-state: controls t 1.20, p 0.24, migraine t 1.88, p 0.079; tran- For Average/Mean Defect, there was a significant inter- 25 ¼ ¼ 15 ¼ ¼ sient: controls t25 1.63, p 0.12, migraine t16 1.58, action between group, visit and perimeter (Huynh-Feldt p 0.13). Neither¼ was there¼ a significant change¼ in " 0.83, F(1.66,68.1) 4.02, p 0.029). The change PVER¼ timing (Table 2; group visit interaction: with¼ time was evident¼ in the control¼ group only – an F(1,40) 0.95, p 0.34), PVER/PERG ratio (Table 2; improvement in SAP Average Defect at the second ¼ ¼ group visit interaction: F(1,40) 0.05, p 0.82), or visit (Table 4; paired t-test: t25 4.80, p < 0.001). The interhemispheric amplitude asymmetry¼ ¼ (Table 3; migraine participants tended to show¼ worse generalised paired Wilcoxon signed rank tests, p > 0.05) with time sensitivity closer to a migraine; that is, a decrease in post-migraine. Although the PERG is normal in TMP Average Defect and an increase in SWAP Mean between migraine attacks (6,8,14), differences in the Defect at the second visit. These changes, however, did PERG may manifest closer to a migraine. We did not not reach statistical significance (Table 4; paired t-tests: find evidence for such an effect (Table 2; group visit p > 0.05). The Pattern Defect/Loss Variance is shown in interactions: p > 0.05).  Figure 3 and was similar at both visits for all participants
302 XML Template (2013) [20.8.2013–3:23pm] [1–16] //blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/CEPJ/Vol00000/130154/APPFile/SG-CEPJ130154.3d (CEP) [PREPRINTER stage]
Nguyen et al. 9
(a) (b) (c) Depressed points Depressed points Depressed points 0.4 Better points Better points Better points
0.2
−0.0
−0.2
−0.4 Proportion of total locations
123 4 5 6 7 8 9 10 11 12 13 14 15 16 17 123 4 5 6 7 8 9 10 11 12 13 14 15 16 17 123 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Migraine participant Migraine participant Migraine participant
Figure 4. Proportion of the total number of visual field locations at the first visit (at least seven days after a migraine) that were identified as depressed (black bars) or better (white bars) relative to the lower 8th percentile and upper 92nd percentile limits of control group performance, respectively, for each individual with migraine. The majority of locations identified as abnormal were depressed, not better. A visual field was considered abnormal if there were at least eight SAP,six TMP or five SWAP locations (horizontal dotted lines) that were identified as depressed. Participants in the migraine with aura group (participants 12–17) are shown to the right of each panel. SAP: standard automated perimetry; TMP: temporal modulation perimetry; SWAP: short-wavelength automated perimetry.
(Table 4; paired Wilcoxon signed rank tests: p > 0.05). controls across most of the sensitivity range. Thus, Thus, overall, perimetric global indices did not change people with migraine showed a significant number of between visits, although the migraine participants did points with reduced sensitivity to begin with (Figure 4) not show the same learning benefits as controls. and which were associated with larger losses closer to a When visual fields of migraine individuals were migraine (Figure 5). assessed using point-wise analysis, the majority of loca- tions that were identified as ‘abnormal’ relative to con- Patterns of visual field loss trol group performance were depressed, not better (Figure 4, black bars). This was evident for all visual Where point-wise comparisons revealed a statistically field tasks, although different people were identified as significant number of depressed points, we investigated abnormal for each test. The total number of depressed whether the pattern of visual field loss involved one or points for a given individual did not change with time both eyes. Data from one migraine participant were post-migraine (Wilcoxon signed rank tests: SAP excluded from this analysis because of a false negative Z 0.05, p 0.96; TMP Z 0.89, p 0.38; SWAP rate >30% in one eye. Both monocular and bilateral Z ¼À0.32, ¼p 0.75). However,¼À individuals¼ with visual field defects were observed in our migraine par- migraine¼À showed¼ point-wise changes in sensitivity that ticipants, although the presence of a bilateral defect does fell outside that predicted from control group test–retest not preclude the possibility of two monocular defects. variability. The proportion of migraine participants Only three people with migraine (19%) showed with a significant number of points across the visual normal results for every visual field task at every visit, field where sensitivity was significantly decreased closer compared with 77% controls. Five people with migraine to the end of a migraine was 41% for SAP, 24% for (31%) demonstrated a repeatable bilateral visual field TMP and 47% for SWAP. These proportions were sig- defect (e.g. Figure 6(a)). None gave a homonymous pat- nificantly different from controls (chi-square, SAP tern respecting the vertical midline. Nevertheless, as the p 0.008, TMP p 0.049, SWAP p 0.003). In con- bilateral visual field loss was diffuse and generalised trast,¼ migraine point-wise¼ sensitivity¼ was not signifi- across the entire field, the majority of cases (80%) satis- cantly improved at the second visit, when chi-square fied our less conservative definition for homonymous tests were corrected for multiple comparisons (SAP deficits, where at least one quadrant was flagged as p 0.24, TMP p 0.14, SWAP p 0.049). To illustrate ‘abnormal’ in both eyes. On the other hand, four different this,¼ Figure 5 shows¼ the sensitivity¼ at the first visit as a migraine participants (25%) showed monocular sensitiv- function of sensitivity at the second visit, pooled across ity loss affecting the same eye at both visits. A further the range of visual field locations. Consistent with pre- three people (19%) showed normal fields at the first visit, vious literature (60), both groups showed increased but developed a monocular field defect closer to a variability for locations with low sensitivity. Whereas, migraine (e.g. Figure 6(b)). Monocular defects ranged on average, the migraine and control groups showed from patchy loss affecting all four quadrants of a single similar upper limits of test–retest performance, the eye, to an arcuate scotoma that crossed the vertical mid- lower limits of the migraine group were below that of line. We interpret these as being of retinal origin.
303 XML Template (2013) [20.8.2013–3:23pm] [1–16] //blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/CEPJ/Vol00000/130154/APPFile/SG-CEPJ130154.3d (CEP) [PREPRINTER stage]
10 Cephalalgia 0(0)
(a) (b) (c) 95% 95% 95% 36 5% 5% 5%
30
24
18
12
Sensitivity at visit 2 (dB) 6
0
0 6 12 18 24 30 36 0 6 12 18 24 30 36 0 6 12 18 24 30 36 Sensitivity at visit 1 (dB) Sensitivity at visit 1 (dB) Sensitivity at visit 1 (dB)
Figure 5. Visual field sensitivity at the second visit plotted as a function of sensitivity at the first visit. The shaded area indicates the 90% CI of test–retest performance for the control group. The 5th and 95th confidence limits for the migraine group are shown as individual symbols. Confidence limits were determined for the range of dB values pooled across all visual field locations. Only sensitivity values appearing at least 20 times were included in the analysis in order to obtain a reasonable estimate of the upper 95% and lower 5% confidence limits. For (a) SAP and (c) SWAP, sensitivity was measured in 1 dB steps, whereas for (b) TMP, sensitivity was measured in 3 dB steps. Consistent with previous literature (60), both groups showed increased variability for locations with low sensitivity. However, the migraine group showed lower limits of visual field sensitivity across the range of sensitivity values. Upper limits were similar between groups. SAP: standard automated perimetry; TMP: temporal modulation perimetry; SWAP: short-wavelength automated perimetry.
(a) 2 days post-migraine 14 days post-migraine
Right eyeLeft eye Right eye Left eye
(b) 1 day post-migraine 56 days post-migraine
Right eyeLeft eye Right eye Left eye
Figure 6. Example SAP visual field defects based on point-wise comparisons with control group performance. Shaded squares indicate depressed points; that is, locations where the sensitivity fell below the lower 8th percentile of control group sensitivity. A SAP visual field was considered abnormal if there were at least eight locations across the visual field that were identified as depressed. (a) Diffuse visual field loss in both eyes of a 20-year-old female with migraine with aura. (b) Right eye visual field defect in a 36-year-old female with migraine with aura, showing monocular inferior arcuate loss one day after migraine (left-hand side). Her visual fields were normal at the first visit 56 days after migraine (right-hand side). SAP: standard automated perimetry. Relationship between abnormal interictal amplitude (Figure 2(a)), Pattern Defect/Loss measures of visual function and migraine Variance (Figure 3) and the number of depressed characteristics points based on point-wise comparisons (Figure 4) remained consistently abnormal during the post- Of the visual field and electrophysiological measures migraine period. Spearman rank correlations between analysed in this study, the steady-state PVER these measures, averaged across both visits, were not
304 XML Template (2013) [20.8.2013–3:24pm] [1–16] //blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/CEPJ/Vol00000/130154/APPFile/SG-CEPJ130154.3d (CEP) [PREPRINTER stage]
Nguyen et al. 11
Table 5. Relationship between steady-state PVER amplitude as transient and steady-state electrophysiological and visual field measures averaged across both visits. responses on the same day, which enables comparison between these approaches for measuring visual process- R p s ing in the same individuals. People with migraine show SAP Pattern defect 0.33 0.19 reduced visual field sensitivity to a range of different À Depressed points 0.18 0.48 stimuli, which is consistent with other psychophysical À TMP Pattern defect 0.07 0.80 evidence that deficits in people with migraine are not Depressed points 0.05 0.84 neural pathway-specific (20,24,25,31,33). On the other SWAP Loss variance 0.28 0.27 hand, the steady-state but not transient response at V1 À was abnormal in the migraine group. Faster flickering Depressed points 0.19 0.45 À stimuli have consistently demonstrated differences in Spearman rank correlations were not significant using a Holm-Bonferroni electrophysiological studies of people with migraine correction for multiple comparisons, starting at p < 0.008. (5,9,61). In some cases, as observed here, a clear separ- PVER: pattern visual evoked response; SAP: standard automated perim- etry; TMP: temporal modulation perimetry; SWAP: short-wavelength ation between migraine and control groups was only automated perimetry. measurable in the steady-state response (14,61). Flicker is known to induce higher metabolic demands and increase blood flow in the brain (62). Disrupted neurovascular coupling in migraine (63) may lead to functional abnormalities that depend on flicker rate. The stimulation rate used in the present study ( 8 Hz) corresponds to the temporal frequency that significant (Table 5), implying that these abnormalities produces a maximal change in cerebrovascular might be caused by different mechanisms. To explore response to a flickering checkerboard pattern as mea- the possibility that abnormal interictal measures on sured by fMRI-BOLD (64). Alternatively, the reduc- electrophysiological or visual field tests related to a par- tion in steady-state PVER amplitude may reflect ticular feature of migraine, correlation coefficients were abnormal visual motion processing, as the major determined between the consistently abnormal visual sources of the steady-state PVER are cortical areas functional measures noted above and the migraine V1 and V5/MT (57). Indeed, there is converging evi- characteristics shown in Table 1. None of the correl- dence for altered function in visual motion processing ations was significant (see Appendix, Table A2). pathways from studies involving transcranial magnetic stimulation (65), structural brain imaging (66) and behavioural measures of global motion integration pro- Discussion cessing (1–3). Consistent with earlier reports (16,19,20,22,24), this The results of this study are consistent with both retinal study demonstrates visual field changes with time post- and cortical visual dysfunction being present in people migraine. However, in this study, sensitivity changes with migraine. On the one hand, our electrophysio- were observed with point-wise comparisons and not logical results, like others (6,8,14), argue for a predom- by comparing the perimetric global indices. The inant cortical anomaly in migraine. Steady-state PVER Pattern Defect/Loss Variance was consistently abnor- amplitudes were consistently reduced (Figure 2(a)), mal, with some migraine individuals showing markedly whereas the PERG was normal, implying no diffuse abnormal values at both visits (Figure 2). It may be that retinal dysfunction (Table 2). On the other hand, our participants were not tested close enough to a both binocular and monocular visual field defects migraine to detect a difference with time. Participants were found (Figure 6). Monocular patterns of visual in the previous study were tested the day following field loss arise from pre-chiasmal or retinal dysfunction. migraine offset, with test sessions lasting no more The homonymous nature of migraine visual aura and than one hour (24). Our participants, however, were the nature of some bilateral field loss are supportive of asked to return within one week of a migraine, as we a cortical origin. Thus, visual field tests suggest the anticipated that it would be more difficult for partici- presence of both cortical and retinal dysfunction in pants to arrange a second visit of three hours duration people with migraine. Our findings suggest that the ret- at short notice. As a result, the average time post- inal defects affect small, localised regions, whereas the migraine at the second visit was not 24 hours, but cortical defects tend to involve larger and more general- three days. The more demanding nature of the long ised regions. test session likely biased the timing of the second visit A novel component of this study was the measure- to a day further away from a headache, and prevented ment of SAP, TMP and SWAP visual fields as well migraine participants from completing the second
305 XML Template (2013) [20.8.2013–3:24pm] [1–16] //blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/CEPJ/Vol00000/130154/APPFile/SG-CEPJ130154.3d (CEP) [PREPRINTER stage]
12 Cephalalgia 0(0)
session the day after an attack, where performance is multifocal techniques, however, do not have the likely to be worse. same spatial resolution as the visual field tests A significant number of locations showed a more employed in this study, which identify sensitivity pronounced reduction in sensitivity one to six days losses in people with migraine at discrete locations after a migraine. The decrease in sensitivity was not using test stimuli of 0.5 and 1.7, although deficits a result of increased variability, as the upper limits of in the peripheral visual field have also been found in test–retest performance in both groups indicate a simi- people with migraine using larger targets of 10 (1). lar number and degree of relatively improving loca- We found that an abnormal electrophysiological tions (Figure 5), consistent with a previous report response did not correlate with visual field perform- (24). Differences in visual field sensitivity immediately ance measured in the same individuals. The difference post-migraine are possibly explained by fatigue or between electrophysiological and visual field tests may poor concentration as a result of anti-migraine medi- be related to the spatial extent of the test targets, as cations or the symptoms of migraine itself. We discussed earlier. In addition, for the most part, elec- endeavoured to minimise post-migraine effects by trophysiological measures obtained at least seven days scheduling test visits at least one day after the offset, post-migraine were not significantly different from and not onset, of all migraine symptoms. Moreover, responses obtained, on average, two to three days the changes in sensitivity were apparent for discrete after an attack, which is consistent with other PVER locations across the visual field, whereas fatigue studies in migraine (10,37). In contrast, visual field would be expected to produce an overall reduction sensitivity was worse closer to a migraine, as has in sensitivity. Differences in Average/Mean Defect in been previously reported (16,19,20,24). Without the migraine group did not fall outside the test–retest having measured visual function at multiple times in variability of control group performance (Table 2). An the migraine cycle (i.e. before, during and after an alternative reason for reduced sensitivity is aversion to attack), our interpretation of the literature to date is the test stimuli (67). This might also explain the reduc- that some visual field defects represent adverse seque- tion in steady-state PVER amplitude. We did not for- lae of migraine, as they are worst in the days following mally measure aversion; however, none of the an attack. The effects of migraine can extend from the participants reported discomfort or voluntarily with- central nervous system to peripheral organs (e.g. to the drew from the study during testing. Furthermore, par- retina), which may explain individual cases of ocular ticipants with a strong aversion to visual stimuli are involvement in migraine (e.g. (28)) and the develop- likely to have excluded themselves from volunteering ment of monocular visual field loss closer to a for a study that explicitly involved extensive visual migraine (e.g. Figure 6(b)). Such retinal changes do testing. not manifest as group differences in the PERG, but Although the mechanism for localised visual field are detected by visual field tests that allow spatial deficits in migraine cannot be ascertained from this localisation. However, abnormal cortical electro- study, it has been suggested that decreased sensitivity physiological responses are generally unchanged after might result from localised vascular events (68). a migraine, but have been reported to differ before and Abnormal peripheral vascular flow and vasospastic during an attack (10,36,37). This suggests that changes tendencies in people with migraine (69,70), particu- in neural activity identified using electrophysiology are larly transient retinal vasospasms occurring during a related to cortical susceptibility to migraine attacks, migraine attack (30), could cause altered perfusion given that the pathogenesis of migraine involves the and increase the risk of focal ischaemic damage to brain (27) and the symptomatology of migraine is lar- the optic nerve head (68,70) and retina (28). gely cortical. However, the steady-state (flicker) PERG was The test–retest results of this study also have impli- normal in our migraine group and did not correlate cations for clinical and research settings where perimet- with localised visual field losses. It is worth noting ric and electrophysiological techniques are used. that the pattern electrophysiological measures used Knowing whether deficits are likely to remain stable in this study involve a large, full-field target and over time, or are a temporary consequence of migraine, are therefore global responses, which are not is important for interpretation of test results. The designed to find small localised losses or investigate potential for change in visual function after migraine the spatial extent of visual dysfunction. Future stu- should be considered, as this will affect the ability to dies may take advantage of multifocal techniques determine abnormality and disease progression in (71), which have the potential to provide more infor- people with migraine in comparison with normal test– mation about localised visual field defects. Current retest variability.
306 XML Template (2013) [20.8.2013–3:24pm] [1–16] //blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/CEPJ/Vol00000/130154/APPFile/SG-CEPJ130154.3d (CEP) [PREPRINTER stage]
Nguyen et al. 13
Clinical implications . We show that both cortical and retinal dysfunction can occur in people who have migraine attacks. In some cases, these appear independent of each other, with visual field changes that appear retinal in origin being variable as a function of time post-migraine. . An abnormal result on an electrophysiological test does not predict whether visual field performance will also be abnormal in people with migraine.
Funding interictal phases in migraine. Acta Neurol Scand 2001; This work was supported by the National Health and 104: 301–307. Medical Research Council [grant number 509208] and 12. Ambrosini A, de Noordhout AM, Sandor PS and Australian Research Council [grant number FT0990930] to Schoenen J. Electrophysiological studies in migraine: A author AMM. BN was supported by the Elizabeth and comprehensive review of their interest and limitations. Vernon Puzey Postgraduate Scholarship from the Faculty Cephalalgia 2003; 23: 13–31. of Science at the University of Melbourne. 13. Coppola G, Parisi V, Fiermonte G, et al. Asymmetric distribution of visual evoked potentials in patients with migraine with aura during the interictal phase. Eur J Conflict of interest Ophthalmol 2007; 17: 828–835. None declared. 14. Nguyen BN, McKendrick AM and Vingrys AJ. Simulta- neous retinal and cortical visually evoked electrophysio- logical responses in between migraine attacks. References Cephalalgia 2012; 32: 896–907. 1. McKendrick AM and Badcock DR. Motion processing 15. Lewis RA, Vijayan N, Watson C, et al. Visual field loss in deficits in migraine. Cephalalgia 2004; 24: 363–372. migraine. Ophthalmology 1989; 96: 321–326. 2. Antal A, Temme J, Nitsche MA, et al. Altered motion 16. Drummond PD and Anderson M. Visual field loss after perception in migraineurs: Evidence for interictal cortical attacks of migraine with aura.Cephalalgia 1992; 12: 349–352. hyperexcitability. Cephalalgia 2005; 25: 788–794. 17. Wakakura M and Ichibe Y. Permanent homonymous 3. Ditchfield JA, McKendrick AM and Badcock DR. Pro- hemianopias following migraine. J Clin Neuroophthalmol cessing of global form and motion in migraineurs. Vision 1992; 12: 198–202. Res 2006; 46: 141–148. 18. De Natale RD, Polimeni D, Narbone MC, et al. 4. Shepherd AJ, Beaumont HM and Hine TJ. Motion pro- Visual field defects in migraine patients. In: Mills RP (ed) cessing deficits in migraine are related to contrast sensi- Perimetry Update 1992/93 Proceedings of the Xth tivity. Cephalalgia 2012; 32: 554–570. International Perimetric Society Meeting. Amsterdam/ 5. Diener HC, Scholz E, Dichgans J, et al. Central effects of New York: Kugler Publications, 1993, pp. 283–284. 19. Sullivan-Mee M and Bowman B. Migraine-related visual- drugs used in migraine prophylaxis evaluated by visual field loss with prolonged recovery. J Am Optom Assoc evoked potentials. Ann Neurol 1989; 25: 125–130. 1997; 68: 377–388. 6. Khalil NM. Investigations of visual function in migraine by 20. McKendrick AM, Vingrys AJ, Badcock DR and visual evoked potentials and visual psychophysical tests. Heywood JT. Visual field losses in subjects with migraine PhD Thesis, University of London, UK, 1991. headaches. Invest Ophthalmol Vis Sci 2000; 41: 7. Tagliati M, Sabbadini M, Bernardi G and Silvestrini M. 1239–1247. Multichannel visual evoked potentials in migraine. 21. C¸ omog˘ lu S, Yarangu¨ meli A, Ko¨ z OG, et al. Glaucoma- Electroencephalogr Clin Neurophysiol 1995; 96: 1–5. tous visual field defects in patients with migraine. 8. Shibata K, Osawa M and Iwata M. Simultaneous record- ing of pattern reversal electroretinograms and visual J Neurol 2003; 250: 201–206. evoked potentials in migraine. Cephalalgia 1997; 17: 22. Goodwin D. Transient complete homonymous hemianopia 742–747. associated with migraine. Optometry 2011; 82: 298–305. 9. Shibata K, Osawa M and Iwata M. Pattern reversal 23. McKendrick AM and Badcock DR. An analysis of the visual evoked potentials in migraine with aura and factors associated with visual field deficits measured with migraine aura without headache. Cephalalgia 1998; 18: flickering stimuli in-between migraine. Cephalalgia 2004; 319–323. 24: 389–397. 10. Judit A, Sandor PS and Schoenen J. Habituation of 24. McKendrick AM and Badcock DR. Decreased visual visual and intensity dependence of auditory evoked cor- field sensitivity measured 1 day, then 1 week, after tical potentials tends to normalize just before and during migraine. Invest Ophthalmol Vis Sci 2004; 45: 1061–1070. the migraine attack. Cephalalgia 2000; 20: 714–719. 25. McKendrick AM, Cioffi GA and Johnson CA. Short- 11. Logi F, Bonfiglio L, Orlandi G, et al. Asymmetric scalp wavelength sensitivity deficits in patients with migraine. distribution of pattern visual evoked potentials during Arch Ophthalmol 2002; 120: 154–161.
307 XML Template (2013) [20.8.2013–3:24pm] [1–16] //blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/CEPJ/Vol00000/130154/APPFile/SG-CEPJ130154.3d (CEP) [PREPRINTER stage]
14 Cephalalgia 0(0)
26. Yenice O, Temel A, Incili B and Tuncer N. Short-wave- 44. Blau JN. Migraine prodromes separated from the aura: length automated perimetry in patients with migraine. Complete migraine. BMJ 1980; 281: 658–660. Graefes Arch Clin Exp Ophthalmol 2006; 244: 589–595. 45. Vingrys A, Demirel S and Kalloniatis M. Multi-dimen- 27. Schwedt TJ and Dodick DW. Advanced neuroimaging of sional color, flicker and increment perimetry. In: Mills migraine. Lancet Neurol 2009; 8: 560–568. RP and Wall M (eds) Perimetry Update 1994/95 28. Gutteridge IF, McDonald RA and Plenderleith JG. Proceedings of the XIth International Perimetric Society Branch retinal artery occlusion during a migraine Meeting. Amsterdam/New York: Kugler Publications, attack. Clin Exp Optom 2007; 90: 371–375. 1994, pp. 159–166. 29. Martinez A, Proupim N and Sanchez M. Retinal nerve 46. Demirel S and Johnson CA. Isolation of short-wave- fibre layer thickness measurements using optical coher- length sensitive mechanisms in normal and glaucomatous ence tomography in migraine patients. Br J Ophthalmol visual field regions. J Glaucoma 2000; 9: 63–73. 2008; 92: 1069–1075. 47. Swanson WH, Sun H, Lee BB and Cao D. Responses of 30. Killer HE, Forrer A and Flammer J. Retinal vasospasm primate retinal ganglion cells to perimetric stimuli. Invest during an attack of migraine. Retina 2003; 23: 253–254. Ophthalmol Vis Sci 2011; 52: 764–771. 31. Coleston DM, Chronicle E, Ruddock KH and Kennard 48. Vingrys AJ and Helfrich KA. The Opticom M-600: A new C. Precortical dysfunction of spatial and temporal visual LED automated perimeter. Clin Exp Optom 1990; 73: 3–17. processing in migraine. J Neurol Neurosurg Psych 1994; 49. Hermann A, Paetzold J, Vonthein R, et al. Age-depen- 57: 1208–1211. dent normative values for differential luminance sensitiv- 32. McKendrick AM, Vingrys AJ, Badcock DR and ity in automated static perimetry using the Octopus 101. Heywood JT. Visual dysfunction between migraine Acta Ophthalmol 2008; 86: 446–455. events. Invest Ophthalmol Vis Sci 2001; 42: 626–633. 50. Weber J and Klimaschka T. Test time and efficiency of 33. McKendrick AM and Badcock DR. Contrast-processing the dynamic strategy in glaucoma perimetry. Ger J dysfunction in both magnocellular and parvocellular Ophthalmol 1995; 4: 25–31. pathways in migraineurs with or without aura. Invest 51. Vingrys AJ and Pianta MJ. Developing a clinical prob- Ophthalmol Vis Sci 2003; 44: 442–448. ability density function for automated perimetry. Aust N 34. Tibber MS and Shepherd AJ. Transient tritanopia in Z J Ophthalmol 1998; 26: S101–S103. migraine: Evidence for a large-field retinal abnormality 52. Bach M, Brigell MG, Hawlina M, et al. ISCEV standard in blue-yellow opponent pathways. Invest Ophthalmol Vis for clinical pattern electroretinography (PERG): 2012 Sci 2006; 47: 5125–5131. update. Doc Ophthalmol 2013; 126: 1–7. 35. McKendrick AM and Sampson GP. Low spatial fre- 53. Odom JV, Bach M, Brigell M, et al. ISCEV standard for quency contrast sensitivity deficits in migraine are not clinical visual evoked potentials. Doc Ophthalmol 2010; visual pathway selective. Cephalalgia 2009; 29: 539–549. 120: 111–119. 36. Sand T and Vingen JV. Visual, long-latency auditory and 54. Di Russo F, Pitzalis S, Spitoni G, et al. Identification of brainstem auditory evoked potentials in migraine: Rela- the neural sources of the pattern-reversal VEP. tion to pattern size, stimulus intensity, sound and light Neuroimage 2005; 24: 874–886. discomfort thresholds and pre-attack state. Cephalalgia 55. King-Smith PE and Kulikowski JJ. Pattern and flicker 2000; 20: 804–820. detection analysed by subthreshold summation. 37. Sand T, Zhitniy N, White LR and Stovner LJ. Visual J Physiol 1975; 249: 519–548. evoked potential latency, amplitude and habituation in 56. Luo X and Frishman LJ. Retinal pathway origins of the migraine: A longitudinal study. Clin Neurophysiol 2008; pattern electroretinogram (PERG). Invest Ophthalmol Vis 119: 1020–1027. Sci 2011; 52: 8571–8584. 38. Siniatchkin M, Reich AL, Shepherd AJ, et al. Peri-ictal 57. Di Russo F, Pitzalis S, Aprile T, et al. Spatiotemporal changes of cortical excitability in children suffering from analysis of the cortical sources of the steady-state visual migraine without aura. Pain 2009; 147: 132–140. evoked potential. Hum Brain Mapp 2007; 28: 323–334. 39. Chen WT, Wang SJ, Fuh JL, et al. Peri-ictal normaliza- 58. Meigen T and Bach M. On the statistical significance of tion of visual cortex excitability in migraine: An MEG electrophysiological steady-state responses. Doc study. Cephalalgia 2009; 29: 1202–1211. Ophthalmol 1999; 98: 207–232. 40. Sakai Y, Dobson C, Diksic M, et al. Sumatriptan nor- 59. Holm S. A simple sequentially rejective multiple test pro- malizes the migraine attack-related increase in brain sero- cedure. Scand J Stat 1979; 6: 65–70. tonin synthesis. Neurology 2008; 70: 431–439. 60. Artes PH, Iwase A, Ohno Y, et al. Properties of perimet- 41. Iester M, Mikelberg FS, Courtright P and Drance SM. ric threshold estimates from Full Threshold, SITA Stan- Correlation between the visual field indices and Heidelberg dard, and SITA Fast strategies. Invest Ophthalmol Vis Sci retina tomograph parameters. J Glaucoma 1997; 6: 78–82. 2002; 43: 2654–2659. 42. Lipton RB, Stewart WF, Sawyer J and Edmeads JG. 61. Marrelli A, Tozzi E, Porto C, et al. Spectral analysis of Clinical utility of an instrument assessing migraine dis- visual potentials evoked by pattern-reversal checkerboard ability: The Migraine Disability Assessment (MIDAS) in juvenile patients with headache. Headache 2001; 41: questionnaire. Headache 2001; 41: 854–861. 792–797. 43. International Headache Society. The international classi- 62. Pastor MA, Artieda J, Arbizu J, et al. Human cerebral fication of headache disorders (2nd edition). Cephalalgia activation during steady-state visual-evoked responses. J 2004; 24: 9–160. Neurosci 2003; 23: 11621–11627.
308 XML Template (2013) [20.8.2013–3:24pm] [1–16] //blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/CEPJ/Vol00000/130154/APPFile/SG-CEPJ130154.3d (CEP) [PREPRINTER stage]
Nguyen et al. 15
63. Zaletel M, Strucl M, Bajrovic FF and Pogacnik T. Cou- 67. Marcus DA and Soso MJ. Migraine and stripe-induced pling between visual evoked cerebral blood flow velocity visual discomfort. Arch Neurol 1989; 46: 1129–1132. responses and visual evoked potentials in migraneurs. 68. Flammer J, Pache M and Resink T. Vasospasm, its role Cephalalgia 2005; 25: 567–574. in the pathogenesis of diseases with particular reference 64. Singh M, Kim S and Kim TS. Correlation between to the eye. Prog Retin Eye Res 2001; 20: 319–349. BOLD-fMRI and EEG signal changes in response to 69. Zahavi I, Chagnac A, Hering R, et al. Prevalence of Ray- visual stimulus frequency in humans. Magn Reson Med naud’s phenomenon in patients with migraine. Arch 2003; 49: 108–114. Intern Med 1984; 144: 742–744. 65. Battelli L, Black KR and Wray SH. Transcranial mag- 70. Gasser P and Meienberg O. Finger microcirculation in netic stimulation of visual area V5 in migraine. Neurology classical migraine. A video-microscopic study of nailfold 2002; 58: 1066–1069. capillaries. Eur Neurol 1991; 31: 168–171. 66. Granziera C, DaSilva AF, Snyder J, et al. Anatomical alter- 71. Hood DC, Bach M, Brigell M, et al. ISCEV standard for ations of the visual motion processing network in migraine clinical multifocal electroretinography (mfERG) (2011 with and without aura. PLoS Med 2006; 3: e402. edition). Doc Ophthalmol 2012; 124: 1–13.
Appendix N N n Pointwise analysis of visual fields Cn 1 À 1 Áð À Þ ð Þ A visual field location was deemed abnormally where depressed if the sensitivity fell below the empirical th lower 8 percentile limit of our control group (n 26) N N! ¼ Cn 2 performance. Similarly, sensitivity was considered sig- ¼ n! N n ! ð Þ nificantly decreased at the second visit if the change in ð À Þ sensitivity fell outside the confidence limits of control and is the probability that an individual point will fall group test-retest change. We subsequently determined below the lower confidence limits of control group per- the number of depressed points required to flag the formance ( =0.04). Visual fields were considered abnor- overall visual field result as abnormal, given the total mal (p<0.05) if there were at least n number of points number of locations tested. Assuming that the thresh- that were identified as depressed (p=0.04 for a single olds at individual locations are independent, the prob- point) within a visual field, or portion of visual field, ability (p) that n points, out of a total N, fall below the consisting of N number of test locations (Table A1). lower confidence limit of control group performance is given by
Table A1. The probability (p < 0.05) that a visual field consists of at least n number of statistically abnormal points (a 0.04), given the total number of test locations (N) (A) across the entire visual field, (B) in the upper quadrants of the visual field, and¼ (C) in the lower quadrants of the visual field. Locations at, immediately above, and below the blindspot were not included.
(A) Number of statistically abnormal points across the entire visual field N n p SAP 101 8 0.030 TMP 73 6 0.045 SWAP 50 5 0.035 (B) Number of statistically abnormal points across the upper quadrants N n p SAP 25 4 0.014 TMP 17 3 0.025 SWAP 12 3 0.010 (C) Number of statistically abnormal points across the lower quadrants N n p SAP 24 4 0.012 TMP 16 3 0.021 SWAP 12 3 0.010
SAP: standard automated perimetry; TMP: temporal modulation perimetry; SWAP: short-wavelength automated perimetry.
309 XML Template (2013) [20.8.2013–3:24pm] [1–16] //blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/CEPJ/Vol00000/130154/APPFile/SG-CEPJ130154.3d (CEP) [PREPRINTER stage]
16 Cephalalgia 0(0)
Table A2. Relationship of abnormal measures of electrophysiology and visual field indices, averaged across both visits, with migraine characteristics. Spearman rank correlation coefficients are provided.
SAP Pattern defect Depressed points
R p R p
Age at first migraine (years) 0.09 0.75 0.04 0.89 À Years of migraine 0.40 0.11 0.22 0.39 Weeks between migraine attacks 0.07 0.79 0.18 0.49 Migraine attacks in past year 0.09 0.74 0.27 0.29 À À Estimated number of lifetime attacks 0.09 0.72 0.15 0.55 À MIDAS questionnaire score (days) 0.26 0.32 0.47 0.06 À À Headache duration (hours) 0.05 0.85 0.03 0.90 À TMP Pattern defect Depressed points R p R p Age at first migraine (years) 0.05 0.84 0.12 0.63 À À Years of migraine 0.16 0.54 0.19 0.46 Weeks between migraine attacks 0.27 0.29 0.28 0.28 À À Migraine attacks in past year 0.20 0.43 0.21 0.43 Estimated number of lifetime attacks 0.27 0.29 0.29 0.26 MIDAS questionnaire score (days) 0.30 0.25 0.16 0.53 Headache duration (hours) 0.19 0.46 0.04 0.89 À SWAP Loss variance Depressed points
R p R p Age at first migraine (years) 0.24 0.35 0.28 0.28 Years of migraine 0.28 0.28 0.14 0.60 Weeks between migraine attacks 0.07 0.78 0.05 0.84 À À Migraine attacks in past year 0.09 0.73 0.04 0.87 Estimated number of lifetime attacks 0.19 0.47 0.07 0.80 MIDAS questionnaire score (days) 0.09 0.72 0.16 0.54 À À Headache duration (hours) 0.15 0.55 0.10 0.70 À PVER Steady-state amplitude Rp
Age at first migraine (years) 0.09 0.73 Years of migraine 0.15 0.55 À Weeks between migraine attacks 0.15 0.57 À Migraine attacks in past year 0.13 0.63 Estimated number of lifetime attacks 0.09 0.73 MIDAS questionnaire score (days) 0.54 0.02 Headache duration (hours) 0.11 0.67
MIDAS: Migraine Disability Assessment Score; PVER: pattern visual evoked response; SAP: standard automated perimetry; TMP: temporal modulation perimetry; SWAP: short-wavelength automated perimetry.
310 9 | Appendix C
This appendix includes supplementary material to Experiment 3 (Chapter 4).
9.1 Calibration of equipment
Like the electrophysiological tests, the CRT monitor used for the perceptual centre- surround task was a Sony G520 Trinitron model, but set at a different resolution and frame rate (600 × 800 pixels, 120 Hz). The CRT monitor was gamma corrected using automated software (vsgDesktop Gamma Version 3.63, Cambridge Research Systems, Kent, UK) connected to an Optical photometer (Cambridge Research Systems, Kent, UK). Calibrations were performed approximately monthly over the course of approximately one year. The CRT comprised three colour phosphor guns: red (1931 CIE chromaticity co-ordinates: x = 0.61, y = 0.34), green (x = 0.28, y = 0.59), and blue (x = 0.14, y = 0.07). A total of 32 readings were averaged for each colour phosphor gun. Over this time period, the maximum luminance of the CRT monitor did not significantly decrease (Figure 9.1).
Figure 9.1 – Maximum luminance of the CRT monitor used for the perceptual centre- surround task over the course of one year. There was no significant change in maximum luminance over time (regression line not shown, R2 = 0.03, slope: p = 0.68).
311 9.2 Modelling of contrast response functions
To simplify the model fits in Experiment 3 (Chapter 4), the following F-tests were conducted to determine whether there was a significant improvement in fitting the control group PVER data (n = 20) with more floated parameters.
Firstly, the value of n has been suggested as 1.4 from human PVER contrast response functions (Busse et al. 2009, Tsai et al. 2011). As the product s multiplied by n determined the slope of the contrast gain, floating both of these parameters € could confound the interpretation of the changes observed. To prevent this € confound, n was fixed to 1.4 from the literature so that only s was floated. The € results from the more complicated model ( n floated) and simpler model ( n= 1.4) were compared with an F-ratio. In all optimisation procedures, constraints were set € € such that all parameters had to return positive values and c50 could not exceed the € € maximum contrast tested (97%). The outcome of this comparison showed that both models fit the control group data well, as indicated by R2 values of 0.97 and 0.98 for € the simpler and more complicated models, respectively. However, the F-test showed no significant improvement in floating the parameter n (F (1, 3) = 0.80, p = 0.44). Thus, the simpler model where n was fixed to 1.4 was adopted.
€ Secondly, R0 was the parameter that described ‘noise’. As we inferred this parameter € by measuring responses at 0% contrast, noise was determined from these data (see Figure 4.7 in Chapter 4). Normalised amplitude at 0% contrast was, on average, 0.07 € for the control group, which was used as the value of R0 in the optimisation process.
This was compared to the more complicated model, where R0 , Rmax , and c50 were floated. Constraints were set as above, with the added constraint that the parameter € n was fixed to 1.4. The two model fits resulted in the same goodness-of-fit (R2 = € € € 0.97). However, if R0 was allowed to vary, the modelling returned significantly varied outcomes with no further benefit of an added parameter to float (F (1, 4) = € 0.005, p = 0.95). Thus, these data provided justification for fixing the parameter R0 € to the group normalised amplitude at 0% contrast.
€ 312 9.3 Representative waveforms
Raw waveforms and the corresponding Fourier spectra for 0%, 18%, and maximum (97%) contrast are shown in Figure 9.2 to illustrate an example control saturating contrast gain response and migraine supersaturating response.
Figure 9.2 – Representative waveforms and corresponding Fourier spectra for Experiment 3, steady-state PVER. (A) Control saturating response where amplitude at low (18%) contrast is similar to amplitude at maximum (97%) contrast. (B) Migraine supersaturating response where amplitude at low (18%) contrast is higher than amplitude at maximum (97%) contrast.
313 9.4 Extra statistical analyses
9.4.1 Comparisons between migraine subgroups
The tables below provide justification for pooling the MO and MA participants in Experiment 3 (Chapter 4), as the two groups performed the same for all PVER and perceptual measures of surround suppression/facilitation.
Table 9.1 – RM-ANOVA for migraine subgroup comparisons of the PVER, normalised to the average control group response. Between factor was group (MO, MA) and within factor was contrast (0, 4, 9, 18, 37, 73, 97%). DF: Degrees of freedom; SS: Sum of squares; MS: Mean square.
PVER amplitude DF (error) SS MS F p-value
Group 1 (18) 0.43 0.43 0.44 0.52
Contrast a 1.95 (35.03) 13.56 6.97 28.11 < 0.001
Contrast x group a 1.95 (35.03) 0.58 0.30 1.21 0.31
PVER phase DF (error) SS MS F p-value
Group 1 (18) 0.02 0.02 0.61 0.44
Contrast b 2.47 (44.54) 1.02 0.41 11.54 < 0.001
Contrast x group b 2.47 (44.54) 0.11 0.05 1.27 0.29 a Mauchly’s test of sphericity: χ2(20)=125.97, p<0.001; Huynh-Feldt correction to degrees of freedom: ε=0.32 b Mauchly’s test of sphericity: χ2(20)=126.05, p<0.001; Huynh-Feldt correction to degrees of freedom: ε=0.41
314 Table 9.2 – RM-ANOVA for migraine subgroup comparisons of suppression ratios. Between factor was group (MO, MA) and within factors were centre contrast (20, 40, 60, 80%) and surround contrast (20, 40, 60, 80%). DF: Degrees of freedom; SS: Sum of squares; MS: Mean square.
DF (error) SS MS F p-value
Group 1 (22) 0.02 0.02 0.22 0.65
Centre contrast a 1.91 (41.96) 4.52 2.37 403.95 < 0.001
Centre contrast x group a 1.91 (41.96) 0.01 0.007 1.16 0.32
Surround contrast b 1.93 (42.34) 1.47 0.76 141.04 < 0.001
Surround contrast x group b 1.93 (42.34) 0.02 0.01 1.76 0.19
Centre contrast x surround 6.09 (133.94) 0.02 0.003 0.79 0.58 contrast x group c a Mauchly’s test of sphericity: χ2(5)=23.63, p<0.001; Huynh-Feldt correction to degrees of freedom: ε=0.64 b Mauchly’s test of sphericity: χ2(5)=32.80, p<0.001; Huynh-Feldt correction to degrees of freedom: ε=0.64 c Mauchly’s test of sphericity: χ2(44)=93.21, p<0.001; Huynh-Feldt correction to degrees of freedom: ε=0.68
9.4.2 Contrast at which maximum response occurred
There was no difference between groups in the proportions of people with maximum PVER amplitude occurred at each of the contrast levels tested (Table 9.3)
Table 9.3 – Chi-square group comparisons of the proportion of people with maximum PVER amplitude at each of the contrast levels tested. DF: Degrees of freedom.
a Contrast (%) Controls Migraine DF Chi-square p-value
0 b 0% 0% - - -
4 b 0% 0% - - -
9 5% 17% 1 1.47 0.22
18 5% 13% 1 0.74 0.39
37 11% 8% 1 0.04 0.85
73 15% 13% 1 0.06 0.81
97 65% 50% 1 1.00 0.32 a p-values represent the significance of chi-square tests of proportions - chi-square analysis could not be conducted when the proportion was equal to zero
315
Minerva Access is the Institutional Repository of The University of Melbourne
Author/s: Nguyen, Bao Ngoc
Title: Vision anomalies in migraine
Date: 2013
Citation: Nguyen, B. N. (2013). Vision anomalies in migraine. PhD thesis, Department of Optometry and Vision Sciences, The University of Melbourne.
Persistent Link: http://hdl.handle.net/11343/38609
File Description: Vision anomalies in migraine