Low Spatial Frequency Contrast Sensitivity Deficits in Migraine are not Visual Pathway Selective

Abstract

Some people who experience migraine demonstrate reduced visual contrast sensitivity that is measurable between migraines. Contrast sensitivity loss to low spatial frequency gratings has been previously attributed to possible impairment of magnocellular pathway function. This study measured contrast sensitivity using low spatial frequency targets (0.25–4 c/deg) where the adaptation aspects of the stimuli were designed to preferentially assess either magnocellular or parvocellular pathway function (steady and pulsed pedestal technique). Twelve people with migraine with measured visual field abnormalities and 17 controls participated. Subjects were tested foveally and at 10° eccentricity. Foveally, there was no significant difference in group mean contrast sensitivity. At 10°, the migraine group demonstrated reduced contrast sensitivity for both the stimuli designed to assess magnocellular and parvocellular function (P < 0.05). The functional deficits measured in this study infer that abnormalities of the low spatial frequency sensitive channels of both pathways contribute to contrast sensitivity deficits in people with migraine.

Introduction

Visual symptoms are commonly associated with migraine; these include visual aura, blur and photophobia. As the visual system is clearly involved in the migraine process, there has been considerable study of visual processing in people who experience migraine. Abnormal performance on cortically processed visual tasks has been well documented, including alterations of visual motion processing (1–3), pattern adaptation and masking (4–6) and orientation discrimination (7), although normal orientation discrimination has also been reported (8).

There is also considerable support in the literature for contrast processing abnormalities in people with migraine. Unlike many of the more complex visual tasks that have been used to study vision in migraine, the neural architecture capable of encoding contrast information arises early in the visual system (9). Lesions of the retina, lateral geniculate nucleus (LGN) or cortical area V1 result in profound impairments of contrast sensitivity in both humans and primates (10–13). In contrast, lesions of the extrastriate visual cortical areas in primates leave contrast sensitivity relatively intact (13). In people with migraine, reductions of contrast sensitivity have been measured both using traditional grating stimuli (14, 15) and also across the visual field using clinical forms of perimetry (e.g. (16–18)). Visual field studies have demonstrated that 20–50% of people who experience migraine can manifest contrast deficit depending on the particular visual field task (16–18). Individuals both with and without aura can have reduced contrast sensitivity (14, 17, 19).

There are several motivations for investigating contrast sensitivity reductions in people who experience migraine. First, many subsequent stages of visual processing are dependent, at least in part, on the contrast of the task, and these successive visual tasks are used to study cortical visual processing in migraine. Understanding the nature of any contrast sensitivity differences between people with migraine and non-headache controls is important to ensure that between-group differences attributed to higher cortical processing do not simply reflect reduced input to cortical centres from impaired earlier stages of visual processing. Second, characterizing the nature of contrast sensitivity reduction in migraine may potentially assist in differential or coexistent diagnosis of ophthalmic conditions in people with migraine. Finally, the study of contrast sensitivity loss associated with migraine might help provide insight into which underlying neural mechanisms are altered in affected individuals.

This latter objective was approached by Benedek et al. (14) through the exploration of the spatial and temporal profile of visual contrast sensitivity loss in people who experience migraine without aura (MoA). The authors found a reduction of contrast sensitivity at low spatial frequencies (less than approximately 4 c/deg) that was present in both photopic and scotopic conditions. Both magnocellular (M) and parvocellular (P) visual pathways are involved in the processing of contrast sensitivity, with M neurons having, on average, optimal responses for lower spatial frequencies than average P neurons (9). Benedek et al. (14) have proposed that their finding of reduced contrast sensitivity to low spatial frequency stimuli supports a greater impairment of M pathway function than P pathway function in migraine. In contrast, a recent study by Yenice et al. (15) found contrast sensitivity reductions in people with migraine across the entire range of spatial frequencies tested (1.5–18 c/deg).

It is not clear why the M visual pathway would be affected more than the P pathway by migraine. The M pathway consists of neurons that are, on average, larger and have faster impulse conduction than P pathway neurons (9), resulting in possibly higher metabolic demand (20). Consequently, it is possible that M neurons may be more susceptible to adverse conditions (e.g. a transient reduction in vascular supply). However, there is considerable overlap in the morphology and physiological responses of neurons contributing to M and P pathways (9). For example, approximately 20% of achromatic LGN P cells have optimal spatial frequencies of < 5 c/deg (21). Consequently, the result of Benedek et al. (14) might be explained by impaired function of low spatial frequency sensitive neurons in both M and P pathways.

Abnormal function of both M and P pathways has been suggested by several previous studies. Coleston et al. (22) demonstrated differences between migraineurs and controls using the background modulation method (BMM) with stimulus parameters designed to preferentially assess either M or P systems. Subsequent study of the BMM methodology suggests that M and P systems are likely to be less isolated by this task than originally proposed (23).

Psychophysical findings consistent with M and P system deficits have also been demonstrated by McKendrick and Badcock(19) using a contrast discrimination task. The stimulus employed in that study was an array of four squares of fixed size, where the subject was required to identify one square that was either incremented or decremented in luminance from the remainder. Contrast discrimination was assessed for a range of different luminances, with the migraine group having elevated contrast discrimination thresholds in the mid-periphery relative to the control group for all conditions tested. Elevated contrast detection thresholds for the migraine group (as measured by the threshold luminance for the detection of a single stimulus square) were also present; however, the contrast detection condition did not permit distinction between likely M and P mechanisms (19). Differences identified in a discrimination paradigm at suprathreshold contrast levels do not necessitate that differences will also be apparent at contrast threshold. Hence, although the study of McKendrick et al. suggests abnormal contrast processing in both M and P pathways in some people with migraine, it does not directly test whether this is the case at contrast threshold, nor whether there is a spatial frequency dependence to the deficit pattern.

Leonova et al. (24) recently described a psychophysical method for assessing contrast sensitivity that purports to enable indirect measurement of achromatic M and P pathway contrast sensitivity functions at low spatial frequencies using grating stimuli. It is important to note that psychophysics is incapable of directly measuring neural function. However, convergent evidence from primate neurophysiology and current understanding of the human visual system suggests that the mechanisms recruited by the technique of Leonova et al. (24) are likely to be M and P pathways. The procedure has been used to study inferred M and P pathway contrast sensitivity separately in a number of ophthalmological conditions including retinitis pigmentosa (25), melanoma-associated retinopathy (26) and glaucoma (27). The technique involves the display of the test stimulus on a luminance pedestal, with detection being biased to one or other of the pathways by different interstimulus adaptation. The advantage of this technique is that it enables functional assessment of the low spatial frequency component of the presumed P pathway, which has previously not been assessable psychophysically. A second advantageous aspect of this technique is that the stimuli designed to favour detection by M or P pathways are identical; it is only the interstimulus adaptation that differs (24). Hence, the procedure avoids differential effects of learning and task complexity when assessing the two pathways.

In this study, we used the technique of Leonova et al. (24) to explore the nature of contrast sensitivity reduction in individuals with migraine, specifically to identify whether there was psychophysical evidence for selective loss of either M or P pathway function measured with low spatial frequency stimuli. It was not the purpose of this study to determine whether contrast sensitivity reductions exist in some migraineurs, as this finding has been previously well established using visual field and grating methodologies. Rather, we chose to explore the nature of contrast sensitivity reduction associated with migraine, hence we studied a selected group of migraineurs with documented reductions in visual field sensitivity.

Methods

Subjects

Seventeen non-headache control subjects (aged 18–35 years) and 12 people with migraine (aged 19–44 years) participated. The mean age of the control group was slightly younger than that of the migraine group (mean of control 26 ± 5 years; mean of migraine 33 ± 6 years, t(27) =−3.4, P < 0.05). Contrast sensitivity for low spatial frequencies is unaffected by this small age difference in a young adult cohort (28). Groups were approximately matched for gender (four of 12 migraine participants and six of 17 controls were male).

The migraine and control cohorts were recruited from a larger study exploring visual performance in people with migraine. As part of the larger study, all subjects were required to meet the following visual and refractive criteria: best visual acuity of 6/7.5 or better, refractive errors of less than ± 5.00 diopters (D) spherical and less than ± 2.00 D astigmatism. The refractive errors of the subset of migraine subjects included in this study ranged from +0.50 to −3.50 D sphere (mean =−0.75 D). All had <−1.50 D astigmatism (mean =−0.25 D). Control subjects were approximately refractively matched [range +0.75 to −2.75 D sphere (mean =−1.00 D), 0 to −1.25 D astigmatism (mean = 0.25 D)]. All subjects underwent a detailed eye examination to ensure the absence of coexistent ocular pathology. This included slit lamp and ophthalmoscopic examination, intraocular pressure measured with applanation tonometry (required to be < 21 mmHg) and scanning laser ophthalmoscopic imaging of the optic disc using the Heidelberg Retinal Tomograph (HRT; Heidelberg Engineering, Heidelberg, Germanyeidelberg, GermanhyH). The optic nerve head topography of all subjects was classified as normal using the Moorfield's Regression Analysis (MRA) tool of the HRTIII software. The MRA has excellent sensitivity and specificity (> 85%) for cross-sectional diagnoses of open-angle glaucoma (29). All subjects additionally had no history of systemic pathology or medications known to affect visual function with the exception of migraine. Subjects were not permitted to be taking ongoing systemic medication known to affect visual processing. Those with migraine were not permitted to be taking ongoing antimigraine pharmacotherapy. All migraine sufferers were required to have migraine symptoms meeting the International Headache Society's criteria (30) for either migraine with aura (MA, four observers) or MoA (eight observers). The migraine features of the tested subjects are summarized in Table 1.

Table 1

Migraine characteristics of the tested subjects

Migraine features	Median	Range
Days since last migraine	16	4–210
Years of migraine	19	7–35
Number of migraines in past year	10	1–20
MIDAS Questionnaire score	4	0–9

MIDAS, Migraine Disability Assessment questionnaire (31), a self-report measure that assesses migraine disability by the number of days that subjects are prevented from participating in day-to-day activities due to headaches.

As part of the larger study, visual fields had been previously assessed using several perimetric techniques including standard automated perimetry and flicker perimetry on the Medmont perimeter (Medmont M700, Auto-Flicker test; Medmont Pty Ltd., Camberwell, Victoria, Australia) and Rarebit perimetry (32). Migraineurs were invited to participate in the present study if they demonstrated a visual field deficit on their most recent flicker perimetry test. This test has been described in detail elsewhere (33) and has been previously shown to be effective at documenting visual field changes associated with migraine (17, 33). For the purposes of the study, a visual field defect was defined as three adjacent abnormal points at P < 0.05 or two adjacent points at P < 0.01 on either the age-normal plot (mean deviation) or hill-of-vision plot (pattern deviation). Points adjacent to the blind spot were not used to meet the inclusion criteria. This visual field classification technique has been used elsewhere for classifying neuro-ophthalmic disorders including migraine (19, 34). Most of the visual field defects were mild, with the range of age deviation scores being −1.43 to −14.38 dB (mean −4.36, s.d. 3.75 dB) and the range of pattern deviation scores being 2.22–10 dB (mean 6.14 dB, s.d. 2.12 dB). Clinical inspection of the visual fields assessed for the larger study showed half of the migraine patients to have strictly monocular visual field abnormalities. For the six migraine patients with binocular deficits, none was homonymous. These visual field findings are consistent with previous studies that have shown binocular homonymous visual field deficits to be rarely represented in a migraine population (e.g. (16–18)).

Participants with migraine were tested a minimum of 4 days post migraine to ensure that performance was not affected by acute effects of migraine and to enable wash-out of medications taken for migraine relief.

Human research ethics approval for the project was provided by the Human Research Ethics Committee (Department of Optometry & Vision Sciences, The University of Melbourne). All participants provided written informed consent and all aspects of the study conformed to the tenets of the Declaration of Helsinki.

Contrast sensitivity measures

The stimuli for the contrast sensitivity task were generated using a ViSaGe system (Cambridge Research Systems Pty, Ltd, Rochester, UK) and presented on a gamma-corrected 21-in monitor (Sony Trinitron G520, frame rate 100 Hz). Participants viewed the monitor from a distance of 50 cm and were supported by a chin and forehead rest. Refractive correction for the working distance was supplied when required.

The stimuli used to assess contrast sensitivity are illustrated in Fig. 1 and were identical to those that we have used previously to explore contrast processing in individuals with glaucoma (27). Test stimuli were Gabor patches (sine wave gratings presented in a Gaussian envelope of 2.66°) that were presented in the centre of an 8° square luminance pedestal of 12.5 cd/m². The study of Leonova et al. (24) explored performance for both increment and decrement pedestals. Although in our case the square was always a luminance decrement, we will use the terminology ‘pedestal’ to maintain consistency with previous literature (24). The background subtended 38.3° × 30.5° of visual angle and was 25 cd/m². Contrast sensitivity was measured for spatial frequencies of 0.25, 0.5, 1, 2 and 4 c/deg.

Figure 1

Schematic of the stimulus. (A) Steady pedestal condition (favours the M-pathway). A pedestal square of 12.5 cd/m² was presented continuously on the 25 cd/m² background. The test stimulus (Gabor) was presented briefly (30 ms), followed by further adaptation to the pedestal. The Gabor was oriented at either 45° or 135° with the orientation being chosen randomly for each trial. (B) Pulsed pedestal condition (favours the P-pathway). The adapting phase consisted of the 25 cd/m² background, whereas the test interval showed the decrement pedestal and the Gabor for 30 ms (oriented at either 45° or 135°).

The steady pedestal condition is illustrated in Fig. 1A. The 12.5 cd/m² pedestal was presented continuously throughout the task. After an initial adaptation period of 1 min, the test stimulus was presented for 30 ms against the luminance pedestal, after which the luminance pedestal remained during the interstimulus interval (1.5 s). Detailed exploration of this task by Leonova et al. (24) has demonstrated that the steady pedestal task results in contrast responses consistent with those described for primate M pathways. The task is thought to favour the M pathway, as adaptation to the pedestal alters the response and gain of neurons that are stimulated by the prolonged pedestal presentation (presumably P cells), leaving the M cells to respond to the brief target presentation.

On each trial, the stimulus was randomly chosen to be presented at either 45° or 135° orientation. The task of the observer was to identify the orientation of the stimulus (a two-alternative forced choice), with responses being recorded via a button box (Cambridge Research Systems; CB6). Thresholds were determined using a three-down, one-up staircase strategy where three sequential correct responses resulted in the contrast of the stimulus being decreased by 20%, whereas every incorrect response resulted in a 20% increase in stimulus contrast. The three-down, one-up staircase converges approximately on the 79% correct response level (35). Each spatial frequency condition was tested separately using two interleaved staircases that terminated after six reversals. The result of each staircase was calculated as the mean of the last four reversals, with the final threshold being the average of the two interleaved staircase outcomes.

Figure 1B illustrates the pulsed pedestal condition. Observers adapted to the background luminance for 1 min prior to commencing the task. The test period (30 ms) involved the simultaneous presentation of both the test stimulus and luminance pedestal. All other aspects of the task and thresholding procedure were identical to those of the steady pedestal. Previous work has demonstrated that the pulsed pedestal condition results in contrast responses consistent with those described for primate P pathways (24). The abrupt onset of the luminance pedestal is designed to saturate the M pathway (24).

One eye of each participant underwent contrast sensitivity assessment. Contrast sensitivity was measured within the central visual field and at a single mid-peripheral location. Figure 2A shows an example visual field and indicates the equivalent area encompassed by the contrast sensitivity test. For the mid-peripheral contrast sensitivity task, the stimulus was placed so that the corner of the pedestal square closest to the fovea was located at 10° on a 45° diagonal line from the fovea. For the migraine subjects, a peripheral quadrant was chosen that included the area of reduced visual field sensitivity. It should be noted that the area of decreased sensitivity did not always encroach directly on the pedestal stimulus placement. Control subjects were tested in matched peripheral locations. Box plots of the visual field sensitivity in each of the tested regions are shown in Fig. 2B,C. Central and peripheral tasks were performed separately. Test order (central vs. peripheral, steady vs. pulsed) was randomized between subjects and balanced between groups. For each subject, all measures were made within a single test session of approximately 1 h duration with rest breaks being taken as required.

Figure 2

(A) Example visual field measured using flicker perimetry from one of the migraine participants. The numbers are raw sensitivity values in decibels. A ‘No’ indicates that the stimulus was not seen at the highest luminance available. The superimposed boxes schematically illustrate the area of visual field encompassed by the contrast sensitivity stimuli. (B,C) The distribution of visual field sensitivities within the tested regions for the control and migraine groups. For each individual, an average visual field sensitivity was determined for each region by averaging the four tested visual field locations (within the boxes in (A), where ‘no’ was given a value of 0). The boxes represent the 25th, 50th and 75th percentiles, and the whiskers show the 10th and 90th percentiles. There was no significant difference in median visual field sensitivity centrally between the migraine and control groups (Mann–Whitney rank sum test, P = 0.67); however, the migraine group had significantly reduced median sensitivity in the mid periphery (Mann–Whitney rank sum test, P < 0.001), consistent with the inclusion criteria for the study.

It should be noted that there is one important difference between our stimuli and those used by Leonova et al. (24): we used a Gabor spatial profile whereas Leonova et al. used a D6 stimulus. D6 stimuli vary in overall size as spatial frequency is changed, but have a fixed spatial frequency bandwidth (i.e. the same number of cycles of the grating are visible for all spatial frequencies; hence the spatial extent of the stimulus is varied, being larger for lower than for higher spatial frequencies). In contrast, our Gabor stimulus had the same envelope size for each spatial frequency, resulting in a variation in the number of visible grating cycles across the range of spatial frequencies tested, but keeping the overall stimulus extent constant. We chose the Gabor profile to ensure that the same visual field area was assessed for stimuli of differing spatial frequency. If localized visual field loss is present, it is possible that smaller stimuli (e.g. higher spatial frequency D6) might be placed entirely outside or inside a scotoma, whereas larger stimuli (lower spatial frequency D6) might encompass both normal and abnormal areas of visual field. Such a situation could result in an apparent spatial frequency-specific contrast sensitivity loss. This problem is avoided by using the Gabor stimuli. Extensive pilot testing demonstrated that similar results are obtained with the Gabor stimuli as D6 stimuli in normal observers (data not shown, but see (27) for data obtained with Gabor stimuli).

Results

Contrast sensitivity measured with the steady (presumed M) and pulsed (presumed P) pedestal tasks

Figure 3 shows contrast sensitivity for the centrally viewed condition. Figure 3A shows data for the control group alone (mean ±s.e.) and confirms that there was marked separation between the thresholds measured with the steady and pulsed conditions. This finding is consistent with previous reports and supports the notion that different mechanisms (presumed to be M and P) are involved in detection for the two tasks (24). Figure 3B,C compare performance of the migraine and control groups and also show group mean (±s.e.) data. Means and s.e. were used to describe the data as the majority of the distributions did not depart from statistical normality (Kolmogrov–Smirnov test, P > 0.05).

Figure 3

Control and migraine group performance for the steady and pulsed pedestal contrast sensitivity tasks when viewed with central fixation. Data are shown as group means ±s.e.m. (A) Control group data only, demonstrating the different functions obtained from the two pedestal conditions. These data are repeated in (B) and (C) for comparison with the migraine group.

Group performance was compared using a repeated measures anova (spss 13; SPSS Inc., Chicago, IL, USA). The within-subjects factors were task (steady or pulsed) and spatial frequency (0.25, 0.5, 1, 2, 4), with group being the between-subjects factor. The sphericity assumption was not met (Mauchly's test of sphericity, P < 0.05), hence the anova was interpreted using corrected degrees of freedom (Greenhouse–Geiser test). As is evident from Fig. 3 and has been shown in previous literature (24), significant within-subjects effects of both task (F_1,27 = 88.64, P < 0.001) and spatial frequency (F_2.6,70.34 = 39.1, P < 0.001) were present. Group performance was not significantly different (F_1,27 = 2.68, P = 0.13) and there was no significant interaction between group and task (P = 0.95) or between group and spatial frequency (P = 0.76).

Figure 4 shows contrast sensitivity for the peripherally viewed stimulus. Figure 4A shows the control group performance and demonstrates that sensitivity was higher for the steady pedestal condition than for the pulsed pedestal condition for spatial frequencies < 1 c/deg. Consequently, the assumption that separate mechanisms are detecting the steady and pulsed pedestal stimuli is only defensible for these lowest spatial frequencies. A repeated measures anova was performed with the same design as that used for the foveal data. Once again, there were significant main effects of both task (F_1,27 = 23.29, P < 0.001) and spatial frequency (F_4,71.2 = 71.29, P < 0.001). There was also a significant interaction between task and spatial frequency (P < 0.001), as is easily visualized in Fig. 4. Figure 4B,C shows that migraine group sensitivity was less than that of controls. This difference in mean group sensitivity was statistically significant (F_1,27 = 7.65, P = 0.01). There was no significant interaction between task and group (P = 0.41), or between spatial frequency and group (P = 0.34); hence, no evidence was found for a spatial frequency selective reduction in contrast sensitivity within the migraine group.

Figure 4

Control and migraine group performance for the steady and pulsed pedestal contrast sensitivity tasks when viewed at 10° eccentricity. Data are shown as group means ±s.e.m. (A) Control group data only, demonstrating the different functions obtained from the two pedestal conditions. These data are repeated in (B) and (C) for comparison with the control group.

Comparison of visual field sensitivity with contrast sensitivity

In order to compare visual field sensitivity with contrast sensitivity performance, we calculated Z-scores for each migraine subject for both visual field and contrast sensitivity measures. Z-scores measure how far an individual's data depart from control mean performance in units of control group s.d. The visual field Z-score was determined from the data shown in Fig. 1B,C. For the contrast sensitivity task, for each individual migraine participant we determined their average contrast sensitivity for the 0.25 and 0.5 conditions (as the mechanisms underlying detection are most clearly separable at these spatial frequencies, and as the difference between migraine and control groups was not found to depend upon spatial frequency). As the distribution of visual field Z-scores departed significantly from that of a normal distribution, Spearman rank order correlations were determined between the contrast sensitivity Z-scores and the visual field Z-scores. No statistically significant relationships were found either centrally or peripherally (P > 0.05).

Comparison of migraine features with contrast sensitivity performance

For the limited and selected migraine sample included in this study, there were no significant correlations (Spearman rank order, all P > 0.05) between any of the measures of contrast sensitivity and migraine features [years of migraine, migraine frequency, estimated number of lifetime attacks (years × frequency), days since last migraine]. A rank-order correlation was used as the data for the migraine features did not meet the assumption of normality necessary for parametric correlation procedures. Inspection of the data revealed the sensitivity of those subjects with and without aura to be intermixed.

Discussion

Several previous studies have suggested that M visual pathway dysfunction is associated with migraine (14, 36). As M neurons are, on average, tuned to lower spatial frequencies than P neurons, the presence of low spatial frequency contrast sensitivity loss in migraine has supported the proposed involvement of M pathways (14). However, a subpopulation of P neurons are also sensitive to low spatial frequency stimuli (21, 37). The pulsed pedestal stimuli used in the present study results in contrast detection data consistent with those of primate P pathways, and hence is believed to enable indirect assessment of this low spatial frequency sensitive component of the P system (24). Our study has demonstrated that the functions of both inferred M and P pathways are altered in migraine, and that low spatial frequency contrast sensitivity loss in migraine should not be considered evidence for selective M involvement.

In the current study, we tested relatively low spatial frequencies (< 4 c/deg) because the steady and pulsed pedestal task only reveal separable contrast sensitivity functions for lower spatial frequencies (24). Consequently, it is not clear from our results whether the function of the inferred P system is altered generally in our subjects, or whether only the low spatial frequency sensitive components are involved. Low spatial frequency sensitive neurons are presumed to be larger cells, which might consequently be more susceptible to adverse conditions than smaller cells. A previous study by Benedek et al. (14) has demonstrated reductions in contrast sensitivity for spatial frequencies of < 4 c/deg, but not for higher spatial frequencies in people with MoA, implying functional impairment of larger cells. In contrast, Yenice et al. (15) showed reduced sensitivity for all spatial frequencies tested, including 12 and 18 c/deg, for which sensitivity will presumably be largely dominated by P response.

A difference between our findings and previous studies exploring grating contrast sensitivity (14, 15) is that we found a statistically significant reduction in contrast sensitivity only when testing outside the fovea. We specifically included participants with some visual field abnormality, in order to explore the nature of contrast sensitivity loss within an area of known contrast processing reduction. Previous studies (14, 15) have not used visual fields as an entry criterion for the study; however, it might be reasonable to expect that participants with known visual field loss might be more, rather than less, likely to manifest a central contrast sensitivity loss. Indeed, our migraine group did have lower mean contrast sensitivity than the control group foveally, although this did not reach statistical significance. The sample size of this study was relatively limited (12 participants) but not dissimilar to that of Benedek et al. (15 participants) (14). An additional factor that might influence central contrast sensitivity is length of migraine history. Khalil reported decreased contrast sensitivity in migraineurs with a history of ≥ 30 years, but normal performance in disease durations of < 10 years (38). Only two of our participants had a migraine history of > 30 years.

An alternative form of contrast processing dysfunction that has been reported in migraineurs is aversion to high-contrast, mid-spatial frequency gratings (approximately 4 c/deg) (39, 40). This subjective symptom has been linked to pattern adaptation anomalies in migraineurs that are considered to have a cortical locus (4, 5). It is worth considering whether the measured reduced contrast sensitivity in migraineurs might relate to aversion or pattern adaptation abnormalities. If present, subjective discomfort might possibly distract from the task, hence elevate thresholds in migraineurs. However, aversion is principally an issue for high-contrast gratings that are displayed for at least several seconds. Given that the stimuli used within our study were presented very briefly (30 ms) and at contrasts close to threshold, it is unlikely that different levels of stimulus aversion between migraine and control groups could explain the findings. It is also not readily apparent why pattern adaptation alterations would result in differential sensitivity reductions across the visual field.

In summary, this study has demonstrated that low spatial frequency contrast sensitivity loss associated with migraine is consistent with reductions in function of both inferred M and P visual pathways. The presence of reduced contrast sensitivity in migraineurs should be considered when designing research exploring complex aspects of visual processing that are reliant on accurate encoding of contrast information. It is also advisable that migraineurs are not included in normative clinical databases of contrast sensitivity performance. The underlying aetiology of migraine-related contrast sensitivity loss is still poorly understood. The current study has demonstrated deficits consistent with the involvement of both M and P pathways; however, it is still to be established whether abnormalities arise cortically, precortically, or involve aberrant feedback from cortical to precortical stages of visual processing. These possibilities are not mutually exclusive and may play different roles in different individuals. Proposed mechanisms must be able to account for non-selective involvement of early visual processing, localized patchy deficits within the visual field, and the presence of non-homonymous monocular visual field deficits in some people with migraine in the presence of normal ocular findings (16, 33, 41).