Sage Journals: Discover world-class research

Abstract

Background/Hypothesis

Migraine affects >1 billion people but its pathophysiology remains poorly understood. Alterations in the trigeminovascular system play an important role. We have compared corneal nerve morphology in patients with migraine to healthy controls.

Methods

Sixty patients with episodic (n = 32) or chronic (n = 28) migraine and 20 age-matched healthy control subjects were studied cross-sectionally. Their migraine characteristics and signs and symptoms of dry eyes were assessed. Manual and automated quantification of corneal nerves was undertaken by corneal confocal microscopy.

Results

In patients with migraine compared to controls, manual corneal nerve fiber density (P < 0.001), branch density (P = 0.015) and length (P < 0.001); and automated corneal nerve fiber density (P < 0.001), branch density (P < 0.001), length (P < 0.001), total branch density (P < 0.001), nerve fiber area (P < 0.001), nerve fiber width (P = 0.045) and fractal dimension (P < 0.001) were lower. Automated corneal nerve fiber density was higher in patients with episodic migraine and aura (P = 0.010); and fractal dimension (P = 0.029) was lower in patients with more headache days in the last three months. Automated corneal nerve fiber density predicted a significant amount of the observed variance in pain intensity (adjusted r² = 0.14, partial r = −0.37, P = 0.004) in patients with migraine.

Conclusions

Corneal confocal microscopy reveals corneal nerve loss in patients with migraine. It may serve as an objective imaging biomarker of neurodegeneration in migraine.

Keywords

Migraine biomarker neurodegeneration corneal confocal microscopy

Introduction

Migraine affects over a billion people worldwide, and according to the Global Burden of Disease study it is responsible for more disability than all other neurological disorders combined (1). It is a complex disorder with heterogeneity in clinical presentation involving multiple cortical, subcortical and brainstem areas that regulate autonomic, emotional, cognitive, and sensory functions (2). Migraine manifests with unilateral headache associated with nausea, vomiting, phonophobia and photophobia, especially in those with severe migraine. Biomarkers for migraine may allow a more accurate diagnosis and facilitate individualized medicine with the development of more effective approaches to prevent or treat migraine (3).

Studies provide support for the key role of sensitization and activation of the trigeminovascular system in the pathogenesis of migraine. Recent studies have shown an increased prevalence of white matter hyperintensities (WMH) in patients with chronic migraine compared to episodic migraine and in those with aura (4) as well as WMH in the trigeminal spinal tract (5). Functional magnetic resonance imaging studies have shown activation of the peripheral and central trigeminal system and altered resting state functional connectivity in key brain areas associated with processing pain in patients with migraine (6). Multiple studies have shown altered brain responses to sensory stimuli and atypical functional connectivity of sensory processing regions in patients with migraine (2). Pain associated with mechanical and chemical irritation of the ocular surface is transmitted to the trigeminal ganglion via mechanical and polymodal nociceptors, which project to the trigeminal brainstem nuclear complex and thalamus. The identification of morphological alterations in the trigeminal nerve that might trigger migraine attacks may provide new targets for treating migraine.

We have pioneered the use of corneal confocal microscopy (CCM) to quantify corneal nerve fiber abnormalities in a range of peripheral neuropathies and central neurodegenerative diseases (7). Three previous studies have used CCM to identify an abnormality of corneal nerves in patients with migraine (8 –10). The first study from the USA undertook CCM in 19 patients with chronic migraine and demonstrated a lower corneal nerve fiber density with no change in corneal nerve branch density or length (8). In a subsequent study from India, corneal nerve branch density, length and area were significantly lower in patients with migraine and photophobia compared to migraine patients without photophobia and controls (9). In the most recent study of 10 patients with episodic migraine from China, corneal nerve fiber density, branch density and length were higher (10). We believe the disparities in reported outcomes reflect the small numbers of patients studied and use of different methods to quantify corneal nerve abnormalities. Our study has addressed some of these disparities by recruiting a larger number of patients to enable comparison between those with episodic and chronic migraine and in relation to the presence and absence of aura, pain location and medication use. Moreover, we have undertaken both manual and automated corneal nerve analysis and reported additional metrics including corneal nerve fiber width, area, and fractals to fully characterize corneal nerve involvement in migraine.

Methods

Study participants and setting

This cross-sectional observational study obtained ethics approval by the Institutional Review Board of Kocaeli University Research and Application Hospital (institutional review board #: GOKAEK-2020/5.03) and all participants gave informed written consent prior to participation. This research adhered to the tenets of the declaration of Helsinki. Sixty patients were recruited from the headache polyclinic at Kocaeli University Research and Application Hospital, Department of Neurology between May 2020, and July 2020. This is the primary analysis of the migraine data. Data from healthy controls constitute a subset of a multi-center study conducted at the University of Manchester, UK and Queensland University of Technology, Australia in 2009–2014 (11). To minimize bias, patient and control images were acquired and quantified using the same protocol.

The headache polyclinic at Kocaeli University Research and Application Hospital has a database of 278 patients diagnosed with migraine by a qualified neurologist according to the International Classification of Headache Disorders-3 (12). To prevent bias, computer-assisted randomization was applied to select n = 60 patients with migraine to participate in the study using Study Randomizer (13). Patients with migraine were aged between 18–45 years (49 women and 11 men) and their diagnosis was re-confirmed at the time of CCM examination. They were subcategorized into episodic (n = 32) or chronic (n = 28) migraine according to the 2018 International Headache Society criteria (14).

Exclusion criteria

Exclusion criteria included a concurrent diagnosis of diabetes, active malignancy, hepatic disease, any other known cause of neuropathy, use of preventive migraine medications (e.g., topiramate, propranolol, amitriptyline, botulinum toxin type A etc.), chronic corneal pathologies, dry eye, history of refractive surgery, ocular surgery and any systemic disease known to affect the cornea such as Fabry’s disease, chronic kidney disease, and Sjogren’s disease. Fasting blood glucose, hemoglobin, erythrocyte sedimentation rate, C-reactive protein, thyroid stimulating hormone, free T3, free T4 and vitamin B₁₂ were assessed.

Anterior eye examination

All participants underwent assessment with the Ocular Surface Disease Index, a validated 12-item questionnaire for the detection of ocular surface irritation symptoms related to dry eye disease. Detailed eyelid and anterior segment slit-lamp examination was performed by an ophthalmologist. No ocular staining was observed after staining with fluorescein sodium. Tear break-up time was assessed. The Schirmer test paper was placed on the lateral 1/3 of the lower fornix, and after five minutes, the wetness on the paper was measured in millimeters.

In vivo corneal confocal microscopy

All participants were scanned with the laser in vivo CCM (Heidelberg Retinal Tomograph III Rostock Corneal Module; Heidelberg Engineering GmbH, Heidelberg, Germany). A certified ophthalmologist (RY) performed all CCM scans using the “section” mode and the approximate scanning time for each eye was 5 minutes.

Manual image analysis

Thirty images per subject on average were acquired from the cornea. Four images of the sub-basal nerve plexus of the right eye were selected based on position, depth, and contrast. The images were analyzed manually by a single examiner (AM) using CCMetrics (RA Malik and MA Dabbah; Imaging Science and Biomedical Engineering, University of Manchester, Manchester, UK). Corneal nerve fiber density (CNFD) (no./mm²), corneal nerve branch density (CNBD) (no./mm²), corneal nerve fiber length (CNFL) (mm/mm²) and the tortuosity coefficient (TC) were quantified as per previously established methodology (15) and reported as average values. Previous studies have shown excellent inter- and intra-observer reproducibility for corneal nerve parameters to quantify neuropathy (7). In the present cross-sectional analysis, CCM image acquisition, extraction, and analysis were each performed by a different examiner who was masked to the underlying participant status.

Automated image analysis

Automated CCMetrics was used to perform fully automated CCM image analysis (ACCMetrics, RA Malik, X Chen, and MA Dabbah; Imaging Science and Biomedical Engineering, University of Manchester, Manchester, UK). Automated analysis harnesses machine learning techniques and provides additional metrics of corneal nerve morphology as outlined below and complements manual analysis. Briefly, for accurate image segmentation enhancement and nerve fiber detection are performed using a dual pattern feature descriptor combined with a neural network classifier to distinguish nerve fibers from the background (artifacts and underlying connective tissue). All the endpoints and branch points of the detected nerve fibers are then extracted and used to create a connectivity map. Automated corneal nerve fiber density (ACNFD) (no./mm²), automated corneal nerve branch density (ACNBD) (no./mm²), automated corneal nerve fiber length (ACNFL) (mm/mm²), automated corneal total branch density (ACTBD) (no./mm²), automated corneal nerve fiber area (ACNFA) (um²/mm²), automated corneal nerve fiber width (ACNFW) (um), automated corneal nerve fiber fractal dimension (ACNFrD) were quantified as per previously established methodology (15). Previous studies have demonstrated a strong agreement between manual and automated measurements for CNFD, CNFL and CNBD (15). The diagnostic performance of automated analysis is comparable to the manual method with the advantage of faster analysis time (∼20 seconds vs. 5 minutes per image) and the acquisition of additional parameters (ACTBD, ACNFA, ACNFW and ACNFrD) (15).

Statistical analysis

Statistical Package for Social Sciences for Windows, version 26 (SPSS statistics, IBM Corp., Armonk, N.Y., USA) was used for statistical analysis and Prism 9 for Mac OS, Version 9.4.1 (GraphPad Software, San Diego, California, USA) was used to generate the figures. Post-hoc power calculation using univariate analysis with CNFD as the dependent variable and patient status as the fixed factor showed that a sample of n = 80 subjects provides 100% power with alpha = 0.05. Patients with migraine were matched to healthy controls at a 3:1 ratio based on age so that the frequency distributions of each group were alike. Patients and controls were not matched for gender, as it is not a significant predictor of corneal nerve morphology (16). Data were confirmed to follow a normal distribution by means of a Shapiro-Wilk test (P > .05). Descriptive statistics were used to provide information about variables in the patient and control groups (frequency %, mean, standard error of mean). Fisher’s exact test was used to assess differences in categorical variables (e.g., sex). An independent samples t-test was used to assess differences between groups and one way analysis of variance (post hoc Bonferroni) was used to assesses differences between more than two subgroups. Correlation analysis (Pearson’s r) was performed to determine statistically significant correlations between variables in the dataset (manual and automated CCM measurements, pain frequency, pain intensity and migraine history). Variables which correlated significantly were used as predictors in a multiple linear regression model adjusted for age and sex to assess the influence of corneal nerve morphology on pain associated with migraine. Continuous data are expressed as (mean ± standard error of mean, P-value). In the present study, no data were missing or excluded from the analysis. For the pain location subgroup analysis patients with undefined pain location were excluded. The reported P-values are two-sided and a P < 0.05 was considered significant.

Data availability statement

All anonymized, individual-level data used in this manuscript are available upon request to the corresponding author.

Results

The detailed demographic and clinical results are presented in Tables 1 and 2. There was no significant difference in age (P = 0.85) between patients with migraine and healthy controls. There was a significantly higher number of females among patients with migraine compared to healthy controls (P = 0.003). The majority of patients had a migraine duration of 11–15 years (26.7%); 15 or less headache days over a three-month period (35.0%); and reported symptoms of photophobia/phonophobia (93.3%) (Table 1). Among the migraine groups patients reported use of non-steroidal anti-inflammatory drugs (n = 27); paracetamol (n = 25); or no medication (n = 8) at the time of assessment. Patients reported pain on the right side (n = 23), left side (n = 32) and 5 were classified as undefined (Table 1). Patients with chronic compared to episodic migraine had a significantly higher number of headache days in the last three months (47.36 ± 2.76 vs 16.66 ± 1.82, P < 0.001) with no significant difference in age (35.07 ± 1.77 vs 34.31 ± 1.45, P = 0.74), migraine duration (14.46 ± 1.63 vs 14.91 ± 1.42, P = 0.84) and pain intensity (7.54 ± 0.23 vs 7.84 ± 0.20, P = 0.31). There were no significant differences in the frequency of visual (p = 0.16), sensory (p = 0.29) or aphasic (p = 0.33) aura between patients with episodic and chronic migraine (Table 2).

Table 1.

Clinical characteristics of patients with migraine, number (%).

	Controls(n = 20)	Patients with Migraine(n = 60)	P-value
Age (years)	35.10 ± 2.21	34.67 ± 1.12	0.85
Sex n (%)
Female	9 (45%)	49 (81.7)	0.003
Male	11 (55%)	11 (18.3)
Migraine Type n (%)
Episodic with aura	0 (0)	25 (41.6)	–
Episodic without aura	0 (0)	7 (11.7)
Chronic	0 (0)	28 (46.7)
Migraine History n (%)
0–5 Years	0 (0)	9 (15.0)	–
6–10 Years	0 (0)	13 (21.7)
11–15 Years	0 (0)	16 (26.7)
16–20 Years	0 (0)	8 (13.3)
>21 Years	0 (0)	14 (23.3)
Photophobia/phonophobia n (%)
No	0 (0)	4 (6.7)	–
Yes	0 (0)	56 (93.3)
Pain Location n (%)
Right side	0 (0)	23 (38.3)	–
Left side	0 (0)	32 (53.4)
Undefined	0 (0)	5 (8.3)
Medication use n (%)
NSAID	0 (0)	27 (45.0)	–
Paracetamol	0 (0)	25 (41.6)
OFF	0 (0)	8 (13.4)

NSAID: non-steroidal anti-inflammatory drug; OFF: no medication at the time of assessment.

Age is expressed as mean ± standard error of mean and all other values are expressed as %.

Table 2.

Frequency of different types of auras in patients with episodic migraine.

Aura Type	Episodic Migraine with Typical Aura (n = 25)	P-value
Visual	n (%)
No	7 (28.0%)	0.16
Yes	18 (72.0%)
Sensory
No	21 (84.0%)	0.29
Yes	4 (16.0 %)
Aphasic
No	22 (87.8%)	0.33
Yes	3 (12.2%)

Anterior eye examination

The Ocular Surface Disease Index score did not differ significantly between females and males with migraine (6.11 ± 1.24 vs 5.94 ± 1.65, p = 0.10); female and male controls (6.31 ± 2.45 vs 6.85 ± 2.86, p = 0.21); and patients with episodic compared to chronic migraine (6.20 ± 1.17 vs 6.29 ± 1.24, P = 0.701). The Schirmer test results did not differ significantly between females and males with migraine [10.521 ± 2.57 vs 11.23 ± 2.67, p = 0.15]; female and males in the control group [11.55 ± 1.96 vs 12.07 ± 2.11, p = 0.18]. Tear break-up time did not differ significantly between females and males with migraine [10.17 ± 1.89 vs 11.03 ± 1.98, p = 0.12]; female and male controls [11.51 ± 2.47 vs 12.17 ± 2.88, p = 0.12].

Manual corneal nerve analysis

Representative images of the sub-basal corneal nerves from a control subject and patients with migraine are presented in Figure 1 and the detailed results are presented in Tables 3–4. Manual CNFD (P < 0.001), CNBD (P = 0.015) and CNFL (P < 0.001) were significantly lower with no difference in TC (P = 0.291) in patients with migraine compared to controls (Table 3, Figure 2). Further subgroup analysis showed that differences remained significant between controls and migraine subtypes for CNFD and CNFL, with no significant difference between patients with episodic or chronic migraine; and between patients with episodic migraine with and without aura (Table 4). There was no significant difference in manual corneal nerve fiber measures for different disease durations (0–5 vs. 6–10 vs. 11–15 vs. 16–20 vs. >21 years); number of headache days in the last three months; and pain intensity. There was no significant difference in manual corneal nerve fiber measures based on pain location and type of medication. TC was significantly lower in patients with visual (15.14 ± 0.38 vs 17.81 ± 0.84, P = 0.034) and speech-language (14.36 ± 0.43 vs 17.06 ± 0.67, P = 0.04) aura compared to patients without aura.

Figure 1.

Representative CCM images from study participants. White arrows indicate main corneal nerve fibers and white arrowheads indicate main corneal nerve branches. Images from an age-matched healthy control (a) and a patient with migraine (b) showing a reduction in (CNFD: for control 43.75 fibers/mm²; for patient 24.99 fibers/mm²) and branches (CNBD: for control: 99.99 branches/mm²; for patient 74.99 branches/mm²) as also indicated by less white arrows and arrowheads in image B compared to A.

Table 3.

Manual and automated CCM measures in healthy controls and patients with migraine.

	Controls(n = 20)	Migraine(n = 60)	P-value
Manual Analysis
CNFD (no./mm²)	39.70 ± 1.00	27.94 ± 0.81	<0.001
CNBD (no./mm²)	105.05 ± 10.0	81.12 ± 4.42	0.02
CNFL (mm/mm²)	28.69 ± 1.01	23.16 ± 0.68	<0.001
TC	14.93 ± 0.79	16.17 ± 0.62	0.29
Automated Analysis
ACNFD (no./mm²)	34.11 ± 1.14	24.94 ± 0.75	<0.001
ACNBD (no./mm²)	49.47 ± 4.16	31.49 ± 1.86	<0.001
ACNFL (mm/mm²)	19.25 ± 0.52	14.10 ± 0.35	<0.001
ACTBD (no./mm²)	68.77 ± 6.00	44.46 ± 2.46	<0.001
ACNFA (mm²/mm²)	0.007 ± 0.0004	0.005 ± 0.0002	<0.001
ACNFW (mm/mm²)	0.02 ± 0.0003	0.02 ± 0.0002	0.05
ACNFrD	1.52 ± 0.005	1.49 ± 0.004	<0.001

Data are expressed as mean ± standard error of mean. CNFD: corneal nerve fiber density; CNBD: corneal nerve branch density; CNFL: corneal nerve fiber length; TC: tortuosity coefficient; ACNFD: automated corneal nerve fiber density; ACNBD: automated corneal nerve branch density; ACNFL: automated corneal nerve fiber length; ACTBD: automated corneal total branch density; ACNFA: automated corneal nerve fiber area; ACNFW: automated corneal nerve fiber width; ACNFrD: automated corneal nerve fractal dimension.

Table 4.

Manual and automated corneal nerve measures in controls, patients with episodic migraine with and without aura, and patients with chronic migraine.

	Controls(n = 20)	Episodic MigraineWithout Aura(n = 7)	Episodic MigraineWith Aura(n = 25)	Chronic Migraine(n = 28)	P-value
Manual Analysis
CNFD (no./mm²)	39.70 ± 1.00	25.92 ± 1.599^A	28.46 ± 1.164^A	28.36 ± 1.370^A	All P < 0.001
CNBD (no./mm²)	105.05 ± 10.0	78.76 ± 9.473	81.44 ± 6.517	81.80 ± 7.406	P = 0.20–1.0
CNFL (mm/mm²)	28.69 ± 1.01	22.96 ± 1.556^A	23.06 ± 1.112^A	23.31 ± 1.066^A	P = 0.003–0.005
TC	14.93 ± 0.79	16.70 ± 1.558	14.72 ± 0.640	17.03 ± 1.051	P = 0.45–1.00
Automated Analysis
ACNFD (no./mm²)	34.11 ± 1.14	21.95 ± 0.97^A	26.45 ± 1.07^A,B	24.99 ± 1.30^A	All P < 0.001;^B,CP = 0.01
ACNBD (no./mm²)	49.47 ± 4.16	33.16 ± 4.25^A	29.16 ± 2.73^A	32.58 ± 3.04^A	P < 0.001–0.04
ACNFL (mm/mm²)	19.25 ± 0.52	13.34 ± 0.62^A	14.31 ± 0.53^A	14.25 ± 0.61^A	All P < 0.001
ACTBD (no./mm²)	68.77 ± 6.00	44.95 ± 5.21^A	43.0 ± 3.71^A	45.37 ± 4.08^A	All P = 0.002–0.02
ACNFA (mm²/mm²)	0.007 ± 0.0004	0.005 ± 0.0004^A	0.005 ± 0.0003^A	0.005 ± 0.0003^A	P < 0.001–0.003
ACNFW (mm/mm²)	0.02 ± 0.0003	0.02 ± 0.0004	0.02 ± 0.0003	0.02 ± 0.0002	P = 0.30–1.0
ACNFrD	1.52 ± 0.005	1.48 ± 0.008^A	1.48 ± 0.006^A	1.49 ± 0.006^A	P < 0.001–0.003

Data are expressed as mean ± standard error of mean; A: significantly different compared to controls; B: significantly different compared to patients with episodic migraine without aura. CNFD: corneal nerve fiber density; CNBD: corneal nerve branch density; CNFL: corneal nerve fiber length; TC: tortuosity coefficient; ACNFD: automated corneal nerve fiber density; ACNBD: automated corneal nerve branch density; ACNFL: automated corneal nerve fiber length; ACTBD: automated corneal total branch density; ACNFA: automated corneal nerve fiber area; ACNFW: automated corneal nerve fiber width; ACNFrD: automated corneal nerve fractal dimension.

Figure 2.

Graphic illustration of manual CCM data. Graphs indicate mean (bold black line) ± range for CNFD (a), CNBD (b), CNFL (c) and TC (d) in controls (left-blue color) compared to migraine (right-red color).

Automated corneal nerve analysis

Automated corneal nerve fiber density (P < 0.001), ACNBD (P < 0.001), ACNFL (P < 0.001), ACTBD (P < 0.001), ACNFA (P < 0.001), ACNFW (P = 0.05) and ACNFrD (P < 0.001) were significantly lower in patients with migraine compared to controls (Table 3, Figure 3). Further subgroup analysis showed that differences remained significant between controls and migraine subtypes for ACNFD, ACNBD, ACNFL, ACTBD, ACNFA, and ACNFrD, with a further significant reduction in ACNFD in patients with episodic migraine without aura compared to patients with aura (P = 0.01). There was no significant difference in automated corneal nerve measures between patients with episodic or chronic migraine (Table 4). There was no significant difference in automated measures of corneal nerve fibers for different disease durations (0–5 vs. 6–10 vs. 11–15 vs. 16–20 vs. >21 years); number of headache days in last three months; and pain intensity. ACNFrD (1.47 ± 0.10 vs 1.50 ± 0.007, P = 0.029) was significantly lower in patients with 31–45 headache days compared to patients with >46 headache days in the last three months. There was no significant difference in automated corneal nerve fiber measures based on pain location and type of medication. A multiple linear regression model with pain intensity as the dependent variable and ACNFA as the independent variable adjusted for age and sex showed that ACNFA predicted a significant amount of the observed variance in pain intensity (r = 0.43, adjusted r² = 0.14, partial r = −0.37, P = 0.004).

Figure 3.

Graphic illustration of automated CCM data. Graphs indicate mean (bold black line) ± range for ACNFD (a), ACNBD (b), ACNFL (c), ACTBD (d), ACNFA (e), ACNFW (f), and ACNFrD (g) in controls (left-blue color) compared to migraine (right-red color).

Discussion

The present study has three main findings. First, we have identified significant corneal sub-basal nerve loss in patients with episodic and chronic migraine, which did not differ by number of headache days, and presence of dry eyes. ACNFA, an overall measure of the amount of nerves, was a significant predictor of pain intensity after adjusting for age and sex. Second, ACNFD was significantly higher in patients with episodic migraine with aura compared to patients without aura with no difference by age, number of headache days and intensity and migraine duration. Additionally, TC was significantly lower in patients with visual or speech-language aura and ACNFrD was significantly lower in patients with more headache days in the last three months.

These data add substantially to the findings of three previous smaller studies (8 –10) showing corneal nerve fiber alterations in patients with episodic or chronic migraine, especially those with photophobia (9). The underlying basis for corneal nerve loss in patients with migraine is not clear. Excitation of trigeminal neurons has been shown to play a central role in the onset of migraine attacks. Primary afferent nociceptive unmyelinated C-fibers and thinly myelinated Aδ fibers arising from the ophthalmic division of the trigeminal ganglion project to peripheral and central sites to form a complex pain matrix (17). A substantial body of evidence suggests that expression of the neuropeptide calcitonin gene-related peptide in the C-fibers of the trigeminovascular system is a principal inducer of migraine through activation of downstream signaling cascades (18). Corneal sub-basal nerves are axonal projections of trigeminal ganglion neurons and are exclusively unmyelinated C-fibers (19). The majority (∼70%) of corneal C-fibers are pain-sensitizing polymodal receptors, which release calcitonin gene-related peptide when stimulated resulting in neurogenic inflammation (20). In this context, it is possible that corneal sub-basal nerve alterations in patients with migraine reflect trigeminal pathology, which is further supported by the finding that pain intensity was associated with ACNFA.

Migraine and dry eye disease are comorbid, and a previous study showed that patients with migraine and corneal nerve loss had symptoms of dry eye based on a dry eye questionnaire (DEQ5) even though objective tests of dry eye, which included the tear break-up time and corneal sensitivity were normal (8). In a subsequent study of patients with migraine, tear break-up time was also normal (9). Indeed, in the present study we also show no difference in symptoms of dry eye or objective tests of tear production in patients with migraine. We therefore cannot attribute corneal nerve degeneration to a change in tear production or the presence of dry eye. Previously corneal nerve loss has been demonstrated in several conditions including diabetic neuropathy, Fabry’s disease, amyloidosis, chronic inflammatory demyelinating neuropathy, systemic lupus erythematosus, sarcoidosis, human immunodeficiency virus, long-COVID and in chemotherapy induced peripheral neuropathy (7,21). Of relevance to the current study, we have also shown corneal nerve loss in patients with burning mouth syndrome (22) and trigeminal neuralgia (23).

Migraine aura has been associated with an increased risk for patent foramen ovale, Parkinson’s disease and a higher risk of ischemic stroke and subclinical ischemic lesions (24). Indeed, available evidence suggest a link between white matter lesions and aura due to microvascular abnormalities increasing the risk for both stroke and aura (25). The prevalence of WMH is increased in the brain (4) and spinal tract (5) of patients with migraine. And we have previously shown that corneal nerve loss is associated with WMH in the brain of patients with ischemic stroke (26). Therefore, one might have anticipated greater corneal nerve loss, but we demonstrate a significantly higher automated corneal nerve fiber density in patients with episodic migraine and aura, which clearly requires further study. While corneal nerve TC is increased in diabetic neuropathy (27), in the present study we show that patients with visual or speech-language aura had a significantly lower TC compared to patients without aura.

There is evidence of activation of the peripheral and central nervous system and altered resting state functional connectivity in key brain areas associated with processing pain in patients with migraine (6). Altered brain responses to sensory stimuli with atypical functional connectivity of sensory processing regions (2) is also recognized in patients with migraine. We and others have previously demonstrated corneal nerve loss in patients with Parkinson’s disease and related it to the severity of motor and autonomic dysfunction, cognitive dysfunction, and altered white matter diffusion in the trigeminal nerve (28). Corneal nerve loss has also been demonstrated in people with mild cognitive impairment and dementia and related to the severity of cognitive dysfunction as well as in multiple sclerosis and amyotrophic lateral sclerosis (7).

Automated corneal nerve fractal dimension analysis is a measure of spatial complexity of the main fibers and branches and as such may detect subtle early pathology. We show a significantly reduced ACNFrD in patients with prolonged pain in the last three months, which agrees with previous findings in painful neuropathies (29,30). Indeed, patients with painful diabetic neuropathy have greater corneal nerve loss compared to patients with painless neuropathy and healthy controls (29). ACNFrD has been shown to be the best performing parameter to identify patients with ocular surface neuropathic pain (30). The trigeminal system is involved in migraine attacks and therefore it is plausible that patients with prolonged pain may have more pronounced alterations in nociceptive corneal axons.

We acknowledge several limitations in our study. First, our ophthalmic imaging protocol assessed only the right eye which may have skewed the results. Although, we have previously shown symmetrical corneal nerve loss in a range of peripheral and central neurological disorders (31,32). Future studies should also investigate inter-eye differences in relation to pain location. Second, we did not assess participants for the presence of white matter hyperintensities and as such we cannot exclude the possibility of confounding. Moreover, comparison to neuroimaging measures would shed light into the relationship between corneal nerve and central pathology. Third, the present study is cross-sectional and longitudinal studies are needed to establish the relationship between corneal nerve involvement and chronic migraine. Fourth, we did not assess the relationship between the time from the last migraine aura attack and corneal innervation. Therefore, the extent to which transient blood flow alterations associated with aura (33) may have impacted corneal nerve morphology is unknown. Our study also has strengths. This is the largest and most detailed study to date to show corneal nerve fiber loss in patients with both episodic and chronic migraine, suggesting that trigeminal afferent nerve damage may play a role in the genesis of migraine. In the present study, patients with migraine did not have signs and symptoms of dry eye disease, reassuring us that dry eye was not a confounder for corneal nerve degeneration (34). We show for the first time that the severity of corneal nerve degeneration was associated with pain intensity. This study suggests that CCM may have utility as an imaging biomarker of neurodegeneration in patients with migraine. Further studies are required to assess the relationship between the severity of corneal nerve loss and disease severity, the impact of white matter pathology and any temporal relationship between the last migraine attack and corneal innervation, especially in patients with migraine aura.

Article highlights

Corneal nerves originating from the trigeminal ganglion can be rapidly and objectively quantified using CCM.

Corneal nerves are reduced in migraine, particularly in patients with episodic migraine without aura, and with more frequent headaches.

Automated image quantification enables, rapid, objective, and unbiased quantification of corneal nerves.

CCM may serve as an imaging biomarker of neurodegeneration in migraine.

Footnotes

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

References

Safiri

Pourfathi

Eagan

, et al. Global, regional, and national burden of migraine in 204 countries and territories, 1990 to 2019. Pain 2022; 163: e293–e309.

Schwedt

Chiang

C-C

Chong

, et al. Functional MRI of migraine. Lancet Neurol 2015; 14: 81–91.

Durham

Papapetropoulos

Biomarkers associated with migraine and their potential role in migraine management. Headache 2013; 53: 1262–1277.

Al-Hashel

Alroughani

Gad

, et al. Risk factors of white matter hyperintensities in migraine patients. BMC Neurol 2022; 22: 1–9.

Henssen

Kluin

Kleerebezem

, et al. White matter changes in the trigeminal spinal tract in chronic migraineurs: an ex vivo study combining ultra–high-field diffusion tensor imaging and polarized light imaging microscopy. Pain 2022; 163: 779–785.

Pan

L-LH

Hsiao

F-J

Chen

W-T

, et al. Resting state electrophysiological cortical activity: a brain signature candidate for patients with migraine. Curr Pain Headache Rep 2022; 26: 289–297.

Petropoulos

Ponirakis

Khan

, et al. Corneal confocal microscopy: ready for prime time. Clin Experiment Optom 2020; 103: 265–277.

Kinard

Smith

Singleton

, et al. Chronic migraine is associated with reduced corneal nerve fiber density and symptoms of dry eye. Headache 2015; 55: 543–549.

Shetty

Deshmukh

Shroff

, et al. Subbasal nerve plexus changes in chronic migraine. Cornea 2018; 37: 72–75.

10.

Shen

Dong

Zhou

, et al. Corneal subbasal nerve plexus changes in patients with episodic migraine: an in vivo confocal microscopy study. J Pain Res 2019; 12: 1489–1495.

11.

Pritchard

Edwards

Dehghani

, et al. Longitudinal assessment of neuropathy in type 1 diabetes using novel ophthalmic markers (LANDMark): study design and baseline characteristics. Diabetes Res Clin Pract 2014; 104: 248–256.

12.

Headache Classification Committee of the International Headache Society (IHS) . The international classification of headache disorders, (beta version). Cephalalgia 2013; 33: 629–808.

13.

Phase Locked Software. Study Randomizer, https://www.studyrandomizer.com/ (2017, accessed 27 February 2022).

14.

Headache Classification Subcommittee of the International Headache Society (IHS). The International Classification of Headache Disorders. Cephalalgia 2004; 24: 9–160.

15.

Petropoulos

Alam

Fadavi

, et al. Rapid automated diagnosis of diabetic peripheral neuropathy with in vivo corneal confocal microscopy. Invest Ophthalmol Vis Sci 2014; 55: 2071–2078.

16.

Tavakoli

Ferdousi

Petropoulos

, et al. Normative values for corneal nerve morphology assessed using corneal confocal microscopy: a multinational normative data set. Diabetes Care 2015; 38: 838–843.

17.

Bertels

Mangutov

Conway

, et al. Migraine and peripheral pain models show differential alterations in neuronal complexity. Headache 2022; 62: 780–791.

18.

Iyengar

Johnson

Ossipov

, et al. CGRP and the trigeminal system in migraine. Headache 2019; 59: 659–681.

19.

Al-Aqaba

Dhillon

Mohammed

, et al. Corneal nerves in health and disease. Prog Retinal Eye Res 2019; 73: 100762.

20.

Belmonte

Luna

Gallar

. CGRP is released by selective stimulation of polymodal nociceptor but not of cold receptor nerve fibers of the cornea. Invest Ophthalmol Vis Sci 2003; 44: 1391.

21.

Petropoulos

Ponirakis

Ferdousi

, et al. Corneal confocal microscopy: a biomarker for diabetic peripheral neuropathy. Clinl Therapeut 2021; 43: 1457–1475.

22.

O'Neill

Marshall

Ferdousi

, et al. Corneal confocal microscopy detects small fiber neuropathy in burning mouth syndrome: a cross-sectional study. J Oral Facial Pain Headache 2019; 33: 337–341.

23.

Lee

Böcking

Holle-Lee

, et al. Corneal confocal microscopy demonstrates corneal nerve loss in patients with trigeminal neuralgia. Front Neurol 2020; 11: 661.

24.

Schürks

Rist

Bigal

, et al. Migraine and cardiovascular disease: systematic review and meta-analysis. BMJ 2009; 339: b3914.

25.

Bashir

Lipton

Ashina

, et al. Migraine and structural changes in the brain: a systematic review and meta-analysis. Neurology 2013; 81: 1260–1268.

26.

Kamran

Khan

Salam

, et al. Cornea: a window to white matter changes in stroke; corneal confocal microscopy a surrogate marker for the presence and severity of white matter hyperintensities in ischemic stroke. J Stroke Cerebrovasc Diseases 2020; 29: 104543.

27.

Kallinikos

Berhanu

O'Donnell

, et al. Corneal nerve tortuosity in diabetic patients with neuropathy. Invest Ophthalmol Vis Sci 2004; 45: 418–422.

28.

Arrigo

Rania

Calamuneri

, et al. Early corneal innervation and trigeminal alterations in Parkinson disease: a pilot study. Cornea 2018; 37: 448–454.

29.

Kalteniece

Ferdousi

Azmi

, et al. Corneal nerve loss is related to the severity of painful diabetic neuropathy. Eur J Neurol 2022; 29: 286–294.

30.

Kundu

Shetty

D'Souza

, et al. A novel combination of corneal confocal microscopy, clinical features and artificial intelligence for evaluation of ocular surface pain. PLoS One 2022; 17: e0277086.

31.

Petropoulos

Fitzgerald

Oakley

, et al. Corneal confocal microscopy demonstrates axonal loss in different courses of multiple sclerosis. Sci Rep 2021; 11: 21688.

32.

Petropoulos

Alam

Fadavi

, et al. Corneal nerve loss detected with corneal confocal microscopy is symmetrical and related to the severity of diabetic polyneuropathy. Diabetes Care 2013; 36: 3646–3651.

33.

Busija

Bari

Domoki

, et al. Mechanisms involved in the cerebrovascular dilator effects of cortical spreading depression. Prog Neurobiol 2008; 86: 379–395.

34.

Cox

Kheirkhah

Aggarwal

, et al. Alterations in corneal nerves in different subtypes of dry eye disease: an in vivo confocal microscopy study. Ocular Surface 2021; 22: 135–142.

Corneal confocal microscopy identifies corneal nerve fiber loss in patients with migraine

Abstract

Background/Hypothesis

Methods

Results

Conclusions

Keywords

Introduction

Methods

Study participants and setting

Exclusion criteria

Anterior eye examination

In vivo corneal confocal microscopy

Manual image analysis

Automated image analysis

Statistical analysis

Data availability statement

Results

Anterior eye examination

Manual corneal nerve analysis

Automated corneal nerve analysis

Discussion

Article highlights

Footnotes

Declaration of conflicting interests

Funding

References