Visual Field Losses in Subjects with Migraine Headaches

Abstract

purpose. To characterize the visual fields of subjects with migraine headaches using static and temporal modulation perimetry.

methods. Sixteen subjects with migraines (15 with aura, 1 without) and 15 nonheadache controls were tested. Perimetry was conducted 7 days after the offset of a headache with both static and temporally modulated targets using the Medmont M-600 automated perimeter (Medmont Pty Ltd., Camberwell, Victoria, Australia). Flicker thresholds were measured using the autoflicker test, which varies flicker rate with eccentricity. A subset of four subjects with migraines (3 with aura, 1 without) had the temporal tuning characteristics of their loss evaluated using fixed temporal frequencies (4, 6, 9, 12, and 16 Hz).

results. Field losses were identified with temporal modulation perimetry in 11 of 16 migraine subjects. The majority of these losses occurred in the presence of normal static thresholds (8/11). The deficits displayed temporal tuning, being greatest for higher temporal frequencies (≥9 Hz). None of the subjects revealed deficits typical of cortical lesions. The migraine-without-aura subject displayed a selective loss to temporally modulated stimuli, which was consistent with the aura group. This defect altered over time, decreasing for 30 to 40 days but remaining, to a smaller extent, for up to 75 days after the headache event.

conclusions. Visual dysfunction that is selective for temporally modulated targets occurs in migraine subjects. The migrainous pattern of dysfunction shares some features with that identified in early stages of glaucoma and raises the possibility for a common precortical vascular involvement in these two conditions.

Flicker sensitivity can be measured across the visual field using temporal modulation perimetry (TMP). This technique involves the determination of detection thresholds for flickering targets and can be performed in one of two ways. The first has the flickering component modulating about a mean luminance. ¹ Alternately, the flickering component of the stimulus may be modulated about some luminance other than the mean. This latter case can be described as being associated with a luminance pedestal or DC offset, terminology that describes the shift of the time-averaged luminance of the spot away from the mean background value. ^{2 3} Both of these methods of achieving flickering stimuli have been shown to be useful in identifying visual loss and in some conditions to be more extensive ^{1 2} and to precede those measured by static perimetry. ^{1 4}  

Although both types of flickering stimuli have proven useful in identifying visual loss, the mechanisms isolated by these stimuli are not identical. Flickering targets presented on luminous pedestals (i.e., with a DC offset) have been shown to isolate a mechanism responsible for fast flicker when the flicker rate is greater than 9 Hz. ^{2 3} This mechanism shows greater spatial summation than that mediating the detection of static or slowly modulating targets (<9 Hz). ⁵ Furthermore, it gives a monotonic adaptational response, ^{3 5} which is not the case for that mediating detection of static or slowly modulating targets (<9 Hz), nor that mediating detection of targets modulating about a mean luminance. ⁶ It has been suggested that this fast flicker mechanism reflects L- and M-cone interactions at a color-opponent stage, ³ although rod–cone interactions also have been proposed. ^{5 7 8}  

There are a number of case reports of cortical visual field losses in individuals with migraine headaches, ^{9 10 11} although the applicability of such findings to the population having migraines is less certain. Studies that have considered larger groups of migraineurs have reported unilateral losses using static targets in a substantial proportion of patients. ^{12 13} This high proportion of unilateral field losses suggests a precortical locus in at least some individuals, which appears contradictory to the cortical locus often considered to mediate the appearance of an aura. ¹⁴  

There are several other suggestions of precortical dysfunction in migraineurs. Anomalous performance in migraineurs has been measured using a novel psychophysical measure known as the background modulation technique. ^{15 16} This method involves measuring detection thresholds for a luminous spot that drifts over either a spatially or a temporally modulated background. ^{15 17 18} This task is thought to reflect precortical processing ^{17 18} ; hence the anomalies measured in migraineurs using this task are suggestive of some form of abnormality early in the visual system. 

Interestingly, the precortical visual anomalies isolated in migraine appear selective for temporally modulated stimuli ^{15 16 19} and have been shown to be more pronounced in the midperipheral visual field. ^{16 19} Abnormal flicker fusion frequencies also have been reported in migraineurs, providing further evidence for an alteration of temporal processing. ²⁰ Hence, the literature is suggestive of a precortical anomaly in the visual processing of temporal stimuli of migraineurs, which may be more pronounced in midperipheral regions of the visual field. 

The proposal that migraineurs may have problems with temporal processing is consistent with the observation that kinetic perimetry also has revealed deficits in these individuals. ²¹ However, because static thresholds were not measured in that study, ²¹ it is unknown whether the deficits were selective for the temporal or motion aspects of the stimulus. Nevertheless, it seems that temporal processing anomalies in migraineurs may be identified using appropriate methods of visual field assessment. Furthermore, as a link has been suggested between migraine and glaucoma, ^{22 23} the possibility for a flicker loss in migraine is made more plausible given the well-documented evidence for temporal processing anomalies in glaucomatous neuropathy. ^{1 24} Such a link may provide a mechanism for the existence of precortical dysfunction in migraine and might arise from vascular causes. ^{25 26}  

In this study we considered the temporal processing of subjects with migraines. We were interested in establishing whether flickering targets were more sensitive than static targets for the detection of visual field loss. We also considered the nature of the loss and whether the deficits were characteristic of cortical or precortical dysfunction. Characterizing visual dysfunction in migraineurs may prove useful in monitoring progression and treatment, in addition to providing further knowledge of the pathophysiological processes underlying this complex neurologic condition. 

Materials and Methods

Ethics approval for the project was provided by the Human Research Ethics Committee of the Department of Optometry and Vision Sciences, University of Melbourne, and the research followed the tenets of the Declaration of Helsinki. Subjects gave written informed consent before participation in the study. The group of subjects detailed in this article has formed the basis of a more extensive study considering the nature of visual loss after migraine, which will be reported elsewhere. 

Subjects

Our study group comprised 15 subjects who had migraine headache with aura (aged 23–35 years) and a single case (aged 25 years) of migraine-without-aura (total of 16 with headache), as well as 15 nonheadache controls (aged 22–31 years). All the migraine-with-aura subjects had visual phenomena as part of their aura. Our reason for concentrating on subjects with aura was because this group shows a patent involvement of their visual system during the headache event. However, a single case of migraine-without-aura also was evaluated because he came from our institutional staff and made himself available for extensive test periods, which allowed us to monitor change over time. Because his data appeared identical with the aura group and his testing provided some useful insights into the longitudinal changes of the loss, we decided to include him in the study. However, wherever comparisons are performed between normal controls and the migraine cohort, he has been excluded from the migraine group. 

Subjects were recruited from The Migraine Clinic at The Royal Melbourne Hospital, Melbourne, Australia, and by written advertisements placed throughout the University of Melbourne. Thirteen of the migraineurs and 11 of the control group were women. All subjects had acuities of 6/7.5 (20/25) or better and refractive errors in the range of 4.00 D sphere and ±2.00 D cylinder. Subjects were required to have normal ocular health as assessed by a mydriatic fundus examination and tonometry (<21 mm Hg). All migraine subjects were examined and diagnosed by a neurologist (JH) as having migraine-with-aura (or without, in one case), which met the criteria of the International Headache Society. ²⁷ Nonheadache controls also underwent ophthalmic and neurologic examinations to exclude the possibility of significant headache or other relevant ocular or neurologic conditions. Subjects were requested to maintain a diary during the study to document the nature and duration of their headaches and the use of any palliative treatments. 

Experimental Procedure

All subjects had both eyes tested using random assignment with the Medmont M-600 automated perimeter (Medmont Pty Ltd., Camberwell, Victoria, Australia). This bowl perimeter uses light-emitting diodes (LEDs) as stimuli (λ_max = 565 nm) and has been described in detail elsewhere. ²⁸ In brief, the bowl luminance is 3.2 candela (cd)/m², with a maximum stimulus luminance of 320 cd/m². The LEDs subtend 0.43° (Goldman size III) and are arranged concentrically at various eccentricities ranging from 1° to 50°. In this study, thresholds were measured at 69 locations in the central visual field, with stimuli lying in rings located at 3°, 6°, 10°, 15°, and 22°. Thresholding was achieved with a 6/3-dB staircase using a stimulus duration of 200 (static) or 800 (flicker) ms. This perimeter has been shown to have a high sensitivity (95%) and specificity (83%) when compared to the Humphrey Field Analyzer (Humphrey Systems, Dublin, CA) in measuring threats to fixation within the central 3° of glaucomatous eyes. ²⁹  

Thresholds for static and flickering targets were determined at the same locations in the central 22° of both eyes for all subjects. Flicker thresholds were measured using the autoflicker test, which varies temporal frequency with eccentricity (18 Hz, 3°; 16 Hz, 6°; 12 Hz, 10° and 15°; and 9 Hz, 22°) to improve the dynamic range of the test. For flicker thresholds, subjects were instructed to respond only if the stimulus was perceived as flickering, twinkling, or shimmering. Such criterion setting has been shown to isolate a mechanism sensitive to fast flicker when the temporal modulation is at least 9 Hz. ² The mechanism isolated by such methods shows different spatial and adaptational properties to one isolated at slower flicker rates. ^{3 5}  

To monitor compliance with the flicker task, near-threshold (threshold+ 9 dB), static stimuli were interspersed with the flickering stimuli during the thresholding process. If the subject responded to a static stimulus, the test was paused and the subject was reinstructed to comply with the flicker percept. A false-positive rate of greater than 20% resulted in the test being abandoned and retesting undertaken; such retesting was only required for one subject. For both static and flicker paradigms, fixational accuracy was checked by a blind-spot monitor. All subjects were instructed in the nature of the tasks and allowed a brief practice trial (2 minutes) before testing with each stimulus. 

Visual field measures were made at least 7 days after the cessation of a migraine headache for the aura group, to remove any influences from medication, fatigue, and nausea that might be associated with the migraine event. The migraine-without-aura subject was measured 24 hours after a migraine and then at approximately weekly intervals to monitor the resolution of his visual field deficit. Regular measurements were possible on this subject for 75 days, at which time he experienced another migraine headache and testing ceased. Fortuitously, visual field data from 4 years before this study and before his first migraine attack were available for this person. 

A subset of three migraine-with-aura subjects and our single individual without aura (a total of 4 subjects), all of whom had well-localized field losses, had additional temporal modulation thresholds measured at fixed temporal frequencies across their visual field. The temporal frequencies of 4, 6, 9, 12, and 16 Hz were used to investigate the tuning of the measured deficits. One of our subjects declined to complete the 16-Hz trial because of fatigue. Thresholding was performed at least 7 days after a further headache for the 3 migraine-with-aura subjects and 21 days after a headache for the migraine-without-aura individual, using the eye with the largest deficit. Temporal frequencies were presented in random order for each subject to minimize order-dependent learning or fatigue effects. 

Statistics

The migraine group was compared to the control group using either parametric or nonparametric statistics, where appropriate. Nonparametric statistics were applied if the distributions were skewed or non-Gaussian (Kolmogrov–Smirnov test, P = 0.05). Group comparisons were achieved using the indices generated by the perimeter (average defect and pattern defect) and by our assessment of the number of abnormal points within each migraine subject’s visual field. At the time of the study, the manufacturer of the perimeter had a limited normative data set for flicker outcomes, so we overcame this limitation by comparing between groups or by developing norms based on our own controls. These measures and their application within this study will be described below. 

The Average Defect.

The average defect is an index calculated by the perimeter. It is determined after comparing the subject’s performance, on a pointwise basis, to an internal database of age-matched norms. In this perimeter, the index is based on data that do not contribute to the determination of the pattern defect (see later) and so it attempts to remove any influence of highly abnormal values from its calculation. Negative values indicate worse than expected performance, whereas positive values indicate a relative improvement across the field. We overcame the normative database limitation by comparing the average defect values of our migraine and control groups directly. Hence, if a bias were present in the database, its effect would be cancelled by making such comparisons. 

The Pattern Defect.

The pattern defect is another index returned by the perimeter. It compares each point to the sensitivity expected from the subject’s calculated Hill of Vision. The calculation is achieved by averaging those points at two eccentricities (e.g., 3° and 22°) that have dB values that lie within an allowable window (4 dB at 3°, 6 dB at 22°) of the maximum value in each ring. These averages must themselves comply with the 95% confidence limits for age-matched norms or another ring is considered. Once two suitable rings are found, the slope (Hill of Vision) is calculated and predictions are made on a point-wise basis as to how each point compares to the expected value. As such, the calculation is not based on age-matched norms but on the person’s own sensitivity profile. This index provides an adequate reflection of the local losses found within the visual field, with negative values indicating better than expected performance and positive values suggesting the presence of localized scotomata. The pattern defect was determined for each migraine and control subject, and the performance of the two groups was compared against each other. 

The Number of Abnormal Points.

The number of statistically abnormal points also was chosen as a method of assessing field loss because it is commonly used in a clinical setting when making a qualitative judgment of an outcome. In this case, rather than rely on Medmont’s point-wise, age-matched normative data, we calculated the 95% populational confidence interval for each point based on our control population, assuming a t-distribution with 14 degrees of freedom. In this determination we used the 2.5-percentile point to provide a conservative estimate of abnormality, given that perimetric data are known to show non-Gaussian characteristics. ³⁰  

Assuming that the thresholds at individual points are independent, then the probability that n points, out of a total of N, fall below the lower confidence limit is returned by ³¹  

\[^{N}\mathrm{C}_{n}\ {\alpha}\ (1-{\alpha})^{N-n}\]

where N was 69, and α is the probability that an individual point will fall outside the control confidence interval (α = 0.025). ^N C _n is the binomial coefficient, ³¹ which is equivalent to N!/[n!(N − n)!], where ! signifies the factorial expansion. The upper and lower bounds of the confidence limits differed for each of the 69 test points. 

Using Equation 1 , visual fields were judged to be abnormally depressed (P ≤ 0.02) if they had 5 or more points below our lower limit (P = 0.025 for a single point) for controls. Similarly, an abnormal hypersensitivity (P ≥ 0.98) was identified by 5 or more points exceeding the upper limit for the control group. 

Results

Perimetric Testing

None of our controls had any defects with static perimetry or TMP in either eye when compared to the normative database provided by the manufacturer. 

The average defect index (dB) is given in Table 1a . There was no significant difference between the migraine-with-aura and control groups for either static (analysis of variance [ANOVA], P = 0.97) or flickering targets (ANOVA, P = 0.25), which indicates a similar level of sensitivity across the entire visual field relative to age-matched expectations. 

The pattern defect index (dB) is given in Table 1b . Although there was no significant difference between the two subject groups for the pattern defect of either eye determined with static perimetry (ANOVA, P = 0.10), the migraine group showed significantly larger pattern defects with flickering targets. The distributions for the pattern defects of the migraine group were skewed, suggesting large local losses within the visual fields of some individuals. Nonparametric analysis revealed a significant departure from the control group for both eyes (Mann–Whitney Rank Sum Test: right eye, P = 0.02; left eye, P = 0.01). 

Assessment of individual points for static outcomes established that 3 of the 15 migraine-with-aura subjects gave fields that had 5 or more abnormally depressed points (P = 0.02). However, of this group, there was no bilateral involvement. On the other hand, probability plots for the TMP outcomes showed that 10 of the 15 migraine-with-aura subjects had 5 or more abnormally depressed points (each with P < 0.025) in at least one eye. Of this group of 10 people, the majority (7/10) gave normal static field results on all other indices, and 7 of the 10 had deficits that were unilateral. None of the three bilateral cases were homonymous in nature. Two of the 10 individuals had significant visual field loss within the central 6° (3 or more of the 20 points within the central 6° depressed, P < 0.01), with the remainder having peripheral deficits only (>6°). One of our subjects showed significantly better thresholds (P > 0.975) than controls for both static and flickering targets (hypersensitivity) in the presence of a normal false-positive rate. It is possible that this finding is a chance event. 

Figure 1 shows a loss typical of the bilateral involvement found with flickering targets. This subject returned normal static thresholds. In Figure 1 (and subsequent figures) Medmont M-600 thresholds are given as dB values. The Medmont dB values are expected to show a 6-dB difference from Humphrey dB values (Humphrey Systems, Dublin, CA) because of configurational differences. ²⁸  

The single sufferer of migraine-without-aura gave losses consistent with the aura group. Figure 2 shows his flicker field thresholds measured in July 1993, 4 years before participation in this study and several months before his first migraine attack. 

Subsequent fields for this same patient are shown in Figure 3 , measured 24 hours after his fourth migraine event (July 1997). Static thresholds were normal in the presence of a marked depression to flickering targets in the superior temporal region of the left eye (Fig. 3 ; 24 of 69 points abnormal, P < 0.0001). The fellow eye (Fig. 3) had a similar but smaller flicker loss in the superior temporal region (7 of 69 points depressed, P < 0.001). 

Temporal Frequency Tuning of Field Loss

The field results imply that a selective defect has occurred in migraineurs to targets that are flickering at high flicker rates (>9 Hz). In this section we test this hypothesis directly by considering the pattern defect index (dB) obtained with fixed temporal frequencies from a subset of four people (3 with aura, 1 without) who had well-defined scotomata (see Materials and Methods). All observers were tested at least 7 days after another headache event, after field testing had confirmed the presence of a well-defined field loss. The nonaura subject was tested 21 days after the initial migraine. Figure 4 shows the pattern defect index (dB) as a function of flicker frequency. Because the pattern defect compares each point in the visual field to the subject’s own expected Hill of Vision, larger pattern defect values demonstrate that either the number of points involved or the depth of the localized defect has increased. From Figure 4 it is apparent that all subjects displayed a loss that was frequency dependent. Three of the subjects revealed a deficit for stimuli modulated at greater than 9 Hz, whereas one revealed a deficit for frequencies greater than 6 Hz. Thresholds measured 21 days after migraine for the observer without aura also showed a similar frequency tuning to that found for the migraine-with-aura group (Fig. 4) , suggesting a commonality to the loss. 

Figure 5 is an example of the field results obtained from one of the aura sufferers as a function of flicker frequency. This same subject is represented by the triangles in Figure 4 . Figure 5 shows the dB values measured at each temporal frequency. The shaded values are those that departed from the subject’s expected Hill of Vision at P < 0.025. Intrasubject Hill of Vision comparisons allow for the generalized decrease in sensitivity found with increasing temporal frequency, which is expected even in normal observers. Z-scores were determined for each person on a point-wise basis, using ring fluctuations as the SD, for all points that departed from the Hill of Vision. Any Z-score greater than 2.30 was considered significant at P < 0.025. ³²  

Resolution of the Defect

The migraine-without-aura subject was tested at regular intervals from 24 hours to 75 days after migraine offset. Figure 6 shows his pattern defect index at each test for both eyes. The right eye data are somewhat variable. The left eye shows improvement over the first 30 to 40 days, with relative stability thereafter. A superior arcuate deficit was consistently present in at least 6 of the 69 points (P < 0.006) measured in the left eye for the entire test period. Testing on this subject was abandoned after 75 days because the subject had a further migraine attack. This case demonstrates that the flicker losses can slowly improve over the initial month after the headache. 

Discussion

The prevalence of static field deficits in migraineurs found in this study (3/15 subjects, 20%) is approximately half that previously observed (approximately 40%). ^{12 13} There are several reasons that might explain our lower prevalence. The average age of our migraine subjects was 27 years, making them younger than those recruited to either of the other two study groups (49 years ¹² and 38 years ¹³ ). If migrainous visual field loss results from cumulative damage after repetitive attacks, it would be expected that older subjects should display more field loss. Furthermore, we made our measurements at least 7 days after the offset of a migraine event, but neither of the previous studies report the testing time after migraine for their subjects. ^{12 13} We have shown that some field losses can change over time, consistent with the data of Drummond and Anderson, ²¹ who report significant improvements in kinetic fields over the first 7 days after migraine. If the subjects of the other studies were tested in close proximity to a migrainous event, it is possible that more deficits may have been identified, leading to the greater prevalence of static field loss. 

Nevertheless, aside from the fewer than expected static losses, we have found that deficits with TMP occur commonly in migraine-with-aura patients. These were demonstrated in 10 of the 15 subjects (67%), with the majority being found in the presence of normal sensitivity to static targets (7/10). This latter finding suggests a selective dysfunction for temporally modulated stimuli. This possibility was confirmed in four of our patients by showing that the defect had temporal tuning, being selective for frequencies greater than 6 to 9 Hz (Fig. 4) . 

Our observation that the flicker loss was not exclusive to migraine-with-aura sufferers but also was found in our single migraine-without-aura subject suggests that both conditions might reflect a common cause. Because the difference between the two headache groups rests in the presence or absence of visual dysfunction involving cortical centers, the commonality of a field loss suggests that it reflects involvement at noncortical regions. This possibility is heightened by the fact that none of our patients gave a homonymous defect. Previous studies that have compared headache group characteristics report either a common deficit ^{13 15} or differing outcomes for the two groups. ³³ Although the small sample size of our work limits our capacity to make broad generalizations, the similarities in the visual losses argue for a common deficit. Obviously, larger samples of migraine-without-aura subjects are needed to define the exact nature of the deficit in this group to consider this issue in greater detail. 

One factor evident from our subject without aura, was the relative improvement in the flickering field found after the initial migraine event. Improvements can occur because of changes in the function of the affected region or because of practice and learning effects. We believe that the improvement found in our subject reflects a real change in performance rather than a learning effect. This observer was a practiced and experienced psychophysical subject before entering the study; hence he was unlikely to show practice-related improvements. Furthermore, the relative stability of the fellow eye over the same time frame (Fig. 6) , the stability of the average defect in the affected eye, and the normal flicker result some 4 years earlier (Fig. 2) are not consistent with the possibility that the improvement could have involved a practice or learning effect. 

Our study sample had a bias to female subjects with migraines (13/16). Because migraine effects approximately 6% of men and 15% to 17% of women, ³⁴ we expected a predominance of female volunteers. Indeed, our sample included a higher percentage of women than expected, based on reported prevalence estimates. Although we attempted to match the control and migraine groups according to both age and gender, this was not a perfect match, inasmuch as the control group included 4 (of 15) men. Nevertheless, we doubt whether gender will have any significant impact on our interpretation because 2 of the 3 male migraineurs showed field loss. 

The exact nature of the field loss due to migraine warrants some consideration because it might help identify the locus of the defect. None of our subjects demonstrated visual dysfunction consistent with a cortical locus (bilateral homonymous deficits) for either static or temporally modulated targets. Hence, our findings indicate that precortical visual dysfunction may be common in migraine sufferers, when measured at least 7 days after the migraine event. The aura of migraine is considered a cortical phenomenon, with measurable changes in cerebral blood flow and metabolism being reported during the migraine event. ^{14 35} This fact may seem inconsistent with our proposal for a precortical locus of field loss. However, the other studies that report visual field involvement in migraineurs also have found a high proportion of individuals with unilateral field deficits. For example, Lewis et al. ¹² found unilateral deficits in 14 of 21 affected individuals. Similarly, De Natale et al. ¹³ report that 10 of 17 cases of field loss had unilateral involvement. This is not to say that cortical deficits do not occur, inasmuch as several cases of homonymous deficits have been reported in the literature. ^{9 10 11} However, both our data and that of others ^{12 13} would suggest that homonymous losses form a minority of cases, with none being found in our sample. 

The cause of the precortical involvement is not clear. Several authors have suggested that subtle vascular anomalies may be present in the peripheral vasculature of noncomplicated migraine subjects. ^{25 36} Support for this proposal has arisen from the observation that subjects with a history of migraine have a higher incidence of low-tension glaucoma than does the general population, ^{22 23} with low-tension glaucoma being thought to have an ischemic etiology. ³⁷ Indeed, many of the visual field deficits observed in our study were present in the peripheral visual field and were similar to the arcuate type of deficits reported for early stages of glaucoma. ¹ It should be noted, however, that all the subjects used in this study were aged less than 36 years, and all had normal optic nerves with normal intraocular pressures. In fact, none of these subjects could have been considered as glaucoma suspects or to have manifest glaucomatous neuropathy. It might be hypothesized, however, that if migraine were to be experienced by these individuals on a regular basis, then associated peripheral vasospasm in the region of the optic nerve or surrounding choroid may eventually lead to structural damage, ^{25 26} which might become evident as a low-tension form of glaucomatous neuropathy. If our proposal were correct, then studying the field losses longitudinally in these people might show some interesting associations with developing nerve head or nerve fiber layer damage. Likewise, studying the regional blood flow surrounding the optic nerve or in the peripapillary area of migrainous subjects may provide greater information regarding the underlying processes. 

Any model used to explain the apparent precortical visual field loss in migraine must account for the time course of resolution of the deficits. We were only able to observe a single migraine-without-aura sufferer longitudinally, who demonstrated gradual improvement in his visual field over a period of 30 to 40 days (Fig. 6) . Drummond and Anderson ²¹ have reported an improvement in visual field performance in migraine-with-aura sufferers measured over the first 7 days after a migraine event, a period that was not tested in this study. A better understanding of the time course of migrainous visual deficits may be important to our understanding of the underlying pathogenesis of such functional loss and provide guidelines for clinical assessment of migrainous patients. 

 Submitted for publication March 24, 1999; revised October 26, 1999; accepted November 30, 1999.

 Commercial relationships policy: C2(AJV), C3(AJV), C5(AJV).

 

Table 1.

 

Average Defect and Pattern Defect

	Static			Flicker
		Control	Migraine	Control	Migraine
a. Average defect
Right eyes	0.89 ± 1.89	0.99 ± 1.72	−1.81 ± 1.64	−3.09 ± 1.84
Left eyes	0.69 ± 1.92	0.92 ± 1.49	−2.32 ± 1.57	−2.70 ± 2.06
b. Pattern defect
Right eyes	1.03 ± 1.05	0.41 ± 0.66	1.54 ± 1.36	4.22 ± 4.20*
Left eyes	1.51 ± 1.67	0.82 ± 1.15	1.77 ± 1.86	5.07 ± 4.13*

 Values are group means ± SD, expressed in decibels.

*  Significantly different from control (P < 0.05).

Figure 1.

 

Visual field thresholds (dB) for a migraine-with-aura subject, aged 27 years, tested 10 days after a migraine event. Shading indicates departure from control group performance at P < 0.025. At the time of testing, this subject had suffered migraine for approximately 5 years with an average of four attacks per year. (a) Right eye thresholds for static targets. (b) Right eye thresholds for temporally modulated targets. (c) Left eye thresholds for static targets. (d) Left eye thresholds for temporally modulated targets.

Figure 2.

 

Temporal modulation perimetry threshold values (dB) for the left eye of the migraine-without-aura subject, aged 25 years, measured in July 1993. These values were obtained 4 years before participation in this study and before his first migraine event.

Figure 3.

 

Visual field results for the migraine-without-aura subject, aged 25 years, measured 24 hours after the cessation of a migraine event. Shading indicates departure from control group performance at P < 0.025. (a) Right eye thresholds for static targets. (b) Right eye thresholds for temporally modulated targets. (c) Left eye thresholds for static targets. (d) Left eye thresholds for temporally modulated targets. This result should be compared to that obtained 4 years previously, before the first headache attack (Fig. 2) . No indicates not seen at maximun intensity.

Figure 4.

 

The pattern defect returned by the perimeter as a function of temporal frequency for three migraine-with-aura subjects (unfilled symbols; circles represent the subject from Fig. 1 ) and the one migraine-without-aura subject (filled symbols; subject from Figs. 2 and 3 ).

Figure 5.

 

Visual field thresholds (dB) for one migraine-with-aura subject measured with temporal frequencies fixed across the entire visual field (4, 6, 9, 12, and 16 Hz). This subject is represented by the triangles in Figure 4 . Shading indicates departure from the subject’s expected Hill of Vision at P < 0.025. No indicates not seen at maximum intensity.

Figure 6.

 

The pattern defect (dB) as returned by the perimeter for the autoflicker program for the migraine-without-aura subject for the right (unfilled symbols) and left (filled symbols) eyes, as a function of the number of days after migraine offset. The horizontal dotted line shows the age-matched normal limit (95%) of performance, as determined for the control group using the autoflicker task.