Abstract
Background: Clinical recognition of vestibular migraine (VM) is still hampered by the lack of consensus diagnostic criteria. The aim of this study is a long-term evaluation of clinical criteria for definite (dVM) and probable (pVM) vestibular migraine.
Methods: We re-assessed 75 patients (67 women, age 24–76 years) with dVM (n = 47) or pVM (n = 28) according to previously published criteria after a mean follow-up of 8.75 ± 1.3 years. Assessment included a comprehensive neurotological clinical examination, pure tone audiometry and caloric testing.
Results: dVM was confirmed in 40 of 47 patients with a prior diagnosis of dVM (85%). Fourteen of 28 patients initially classified as pVM met criteria for dVM (50%), nine for pVM (32%). Six additional patients with dVM and two with pVM had developed mild sensorineural hearing loss, formally fulfilling criteria for bilateral Menière’s disease (MD), but had clinical features atypical of MD. Seven of these also met criteria for dVM at follow-up. The initial diagnosis was completely revised for four patients.
Conclusion: Although VM diagnosis lacks a gold standard for evaluation of diagnostic criteria, repeated comprehensive neurotological evaluation after a long follow-up period indicates not only high reliability but also high validity of presented clinical criteria (positive predictive value 85%). Half of patients with pVM evolve to meet criteria for dVM. However, in a subgroup of VM patients with hearing loss, criteria for dVM and MD are not sufficiently discriminative.
Introduction
Vestibular migraine (VM) has been increasingly recognized as a frequent cause of episodic vertigo. However, clinical recognition is still hampered by the lack of generally accepted diagnostic criteria and differential diagnosis from other vestibular disorders such as Menière’s disease (MD) and benign paroxysmal positional vertigo (BPPV) may be difficult because of an overlap of symptoms (1,2).
In the absence of specific neurotological findings, diagnosis of VM is based on a history of recurrent episodes of vertigo unexplained by other central or otological disease in patients with migraine. However, most patients with VM may not be classified according to the current International Classification of Headache Disorders (ICHD) (3). Duration of vertigo attacks does not generally correspond to a typical migraine aura (4,5), only a minority of patients with VM meet criteria for basilar type migraine (5–9) and the link with migraine may be obscured when vertigo occurs dissociated from headache (4–6,10).
Based on these observations, our group has formulated diagnostic criteria for VM that intend to account for the specific characteristics of vertigo in migraine. They provide a specific category of definite VM (dVM) and a tentative category of probable VM (pVM) for those patients who may not exhibit all typical features when VM first manifests (10).
The purpose of this study was to evaluate the validity of clinical criteria for VM by long-term follow-up of patients with a diagnosis of dVM or pVM. We also wanted to test the hypothesis that patients classified, as pVM may evolve to match the stricter criteria of dVM.
Subjects and methods
We re-assessed 75 consecutive patients who had been assigned a diagnosis of definite (dVM) or probable (pVM) VM according to initial neurotological evaluation at our outpatient dizziness clinic after 5–11 years. Diagnosis of dVM and pVM was based on the following criteria (10):
dVM
≥2 attacks of vestibular vertigo (rotational vertigo, other illusory self or object motion, positional vertigo, head motion intolerance, i.e. sensation of imbalance or illusory self or object motion that is provoked by head motion) migraine according to ICHD (3) concomitant migrainous symptoms during ≥2 vertigo attacks no evidence of other central or otological causes of vertigo.
pVM
≥2 attacks of vestibular vertigo at least one of the following:
migraine according to ICHD at least one migrainous symptom during ≥2 vertigo attacks: migrainous headaches photophobia phonophobia visual or other auras no evidence of other central or otological causes of vertigo.
Differing from our previously proposed criteria (10), response to antimigrainous medication or migraine precipitants of vertigo were not included in a definition for pVM in this study, because they were considered less specific indicators for VM.
Exclusion criteria
According to our criteria, we excluded patients with other otological or central vestibular disorders at the time of initial presentation such as previous ear surgery, retrocochlear neoplasm, MD, BPPV, vestibular paroxysms, autoimmune inner ear disease, otosclerosis, otosyphilis, posterior circulation stroke or transitory ischemic attack.
Patients with auditory symptoms were excluded if they fulfilled criteria for MD according to the guidelines of the American Academy of Otolaryngology (AAO) (11):
≥2 episodes of spontaneous rotational vertigo lasting ≥20 min. Tinnitus or aural fullness in affected ear. Audiometrically documented hearing loss on at least one occasion:
The average mean (arithmetic mean) of hearing thresholds at 0.25, 0.5 and 1 kHz is 15 dB or more higher than the average of 1, 2 and 3 kHz or
In unilateral cases, the average of threshold values at 0.5, 1, 2 and 3 kHz is 20 dB or more poorer in the ear in question than on the opposite side or
In bilateral cases, the average threshold values at 0.5, 1, 2 and 3 kHz is greater than 25 dB in the studied ear.
Patients with positional vertigo were excluded if they had positional nystagmus typical of BPPV or if a history of recurrent episodes of purely positional vertigo of ≤1 minute duration was highly suggestive of BPPV. Patients with positional vertigo with atypical features such as atypical nystagmus, duration of vertigo of >1 minute or additional spontaneous vertigo were not excluded. Patients suffering from concomitant vertigo disorders were also excluded. Patients with a concomitant anxiety disorder were excluded if anxiety symptoms were the predominant complaint or could not reliably be distinguished from vestibular symptoms. Patients with insufficient German language command as required for a detailed interview were also excluded.
Clinical characteristics of 75 patients with an initial diagnosis of definite (dVM) or probable (pVM) vestibular migraine at initial presentation
The classification to one of the migraine syndromes as defined by the ICHD was mutually exclusive. If patients met diagnostic criteria for both migraine with and without aura, they were classified as migraine with aura. If they met diagnostic criteria for migraine with aura and basilar migraine, they were classified as basilar migraine. Vertigo was not considered as a migraine aura.
If the ICHD criteria for migraine were missed by one criterion, the patient was classified as probable migraine.
More than one answer possible.
Neurotological assessment at initial presentation and follow-up included a detailed interview covering clinical features such as type, duration, frequency, severity and precipitants of vertigo (i.e. positional change, head movements or situational triggers), a history of migraine according to ICHD, concomitant cochlear and migrainous symptoms with vertigo and a clinical neurotological examination of vestibular and oculomotor function including positional testing. At initial presentation, complementary diagnostic procedures including cerebral imaging, bithermal caloric testing and pure tone audiometry (PTA) were performed to exclude other causes of vertigo on an individual basis.
Additional enquiry at follow-up included a family history of migraine as well as a short-form of the self-administered PRIME-MD Patient Health Questionnaire (PHQ) to diagnose active depressive and panic disorders based on the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV). The PRIME-MD PHQ has been validated to detect any mood disorder with a sensitivity of 61% and specificity of 94%, major depression with 73% and 98% and panic disorders with 81% and 99% compared with diagnoses made by mental health professionals (12). In all patients PTA was performed. Bithermal caloric testing was done in 53 patients (71%). Twenty-two patients refused caloric testing.
At follow-up, diagnosis of VM was re-evaluated based on history and neurotologic findings. If there was evidence of another vestibular or non-vestibular disorder as the cause for vertigo, the initial diagnosis was revised. If an alternative disorder could not fully be excluded, but VM was still the likely diagnosis, it was classified as a competing diagnosis. In patients without evidence of other causes for vertigo, vertigo symptoms were again classified as dVM or pVM according to our criteria, taking into account all available history.
The study was approved by the local ethics board. All participants gave informed consent before entering the study.
Statistical analysis
For statistical analyses SPSS, version 18.0, was used. The chi-squared test or Fisher’s test were applied to test for differences of nominal values. Student’s t-test was used to test for differences in means. A two-sided significance level of p < 0.05 was adopted for all tests. In the absence of a biological marker for VM, validity of clinical criteria for dVM was estimated by calculating a positive predictive value (PPV) for an initial diagnosis of dVM according to our criteria based on a diagnosis of dVM at follow-up as diagnostic gold standard.
Results
At follow-up, a diagnosis of dVM was confirmed in 40 of 47 patients with a prior diagnosis of dVM (85%, Figure 1). Six patients (13%) had cochlear symptoms and had developed mild bilateral sensorineural hearing loss that included the low-frequency range and formally fulfilled AAO criteria for bilateral MD (mean hearing level at 0.5–3 kHz >25 dB in both ears, Figure 2). However, clinical presentation was atypical of classical MD and VM was considered the likely diagnosis. One patient (2%) reported only non-vestibular dizziness due to an anxiety disorder. Considering all information available, it was uncertain whether initial vertigo complaints had truly been of vestibular origin and diagnosis of VM was revised. The positive predictive value (PPV) of criteria for dVM was 85% when based on a final diagnosis of dVM as gold standard. If the six patients with hearing loss are included, PPV rises to 98%.
Follow-up diagnoses of patients with an initial diagnosis of definite vestibular migraine (dVM) and probable vestibular migraine (pVM). Examples of hearing thresholds in four patients with sensorineural hearing loss matching criteria for bilateral Menière’s disease at follow-up. A female, age 68 years, follow-up diagnosis dVM; B female, age 62 years, follow-up diagnosis dVM; C female, age 60 years, follow-up diagnosis dVM; D female, age 65 years, follow-up diagnosis pVM.
Out of 28 participants with a prior diagnosis of pVM, 14 patients (50%) fulfilled criteria for dVM and nine (32%) were still classified as pVM (Figure 1). Two patients (7%) with cochlear symptoms had developed bilateral hearing loss matching AAO criteria for MD. Because of atypical clinical features, however, VM was the likely diagnosis.
In three patients (11%), the initial diagnosis of pVM was revised. One patient with recurrent episodes of purely positional vertigo of ≤1 minute duration, but initially atypical positional nystagmus, showed positional nystagmus typical of posterior canal BPPV at follow-up and a final diagnosis of BPPV was made. One patient, who had initially presented with episodic imbalance and oscillopsia, had developed progressive cerebellar ataxia with spontaneous downbeat nystagmus and no longer experienced vertigo attacks. After exclusion of other causes and based on a positive family history, spinocerebellar ataxia was diagnosed. In a third patient, non-vestibular vertigo due to an anxiety disorder was diagnosed.
Clinical characteristics of patients meeting criteria for MD
All eight patients (seven women) meeting AAO criteria for MD showed clinical features atypical of classical MD. Hearing loss was symmetrical and unusually mild for long-standing MD. Hearing impairment was mild in six (mean hearing level 0.5–3 kHz ≤35 dB) and moderate in two patients (mean hearing level 0.5–3 kHz ≤45 dB; Figure 2). All patients had vertigo attacks atypical for MD such as attacks of long duration (≥3 days, max. duration 42 days; n = 4) or episodes of imbalance with head motion intolerance or recurrent attacks of short-lasting spontaneous or positional vertigo (n = 8). All eight patients had migraine with aura and seven fulfilled criteria for dVM except that another vestibular disorder could not fully be excluded. The median age of onset of vertigo was 46 years (range 14–66 years). The median duration of vertigo at initial presentation was 10 years (range 8 months to 44 years).
Clinical characteristics of patients with confirmed dVM or pVM
Fifty-four of 63 patients with a final diagnosis of VM and no competing diagnosis fulfilled criteria for dVM (86%). Nine patients (14%) had no migrainous symptoms with vertigo and met criteria for pVM. Of the 63 VM patients, 57% had migraine with aura, 43% migraine without aura. Only one patient met criteria for basilar type migraine. The proportion of migraine with aura was significantly higher in patients meeting criteria for dVM (n = 54) compared with nine patients classified as pVM (63% vs. 22%, p = 0.022). There were no significant differences regarding age, sex, age of onset of vertigo, concomitant cochlear symptoms or family history of migraine or vertigo between patients with dVM or pVM.
Concomitant vestibular or psychiatric disorders
In 18 of the 63 patients with a dVM or pVM at follow-up, a concurrent vestibular or psychiatric disorder that could clearly be distinguished from VM was diagnosed. One patient had concurrent BPPV. In 17 patients (27%), at least one concomitant psychiatric disorder was diagnosed. Eleven patients (17%) had psychogenic dizziness. In ten patients (16%), an active depressive (n = 9) or panic disorder (n = 4) according to the PHQ was present.
Discussion
Although it has been widely accepted that recurrent vertigo may be a symptom of migraine, there is still no general agreement on diagnostic criteria. This study shows that our diagnostic criteria of dVM are a valid diagnostic tool in patients with presumed VM. In 85% of patients initially diagnosed as dVM, a diagnosis of dVM was confirmed and no other cause of vertigo was revealed at re-evaluation. Half of patients initially classified as pVM evolved to meet stricter criteria for dVM and an additional 32% remained pVM. In our series, the main differential diagnosis for VM was MD. Six patients with an initial diagnosis of dVM and two with pVM had cochlear symptoms and sensorineural hearing loss meeting AAO criteria for bilateral MD. However, clinical features were atypical of classical MD in all eight patients. Hearing impairment was symmetrical and unusually mild for long-standing MD, in which cochlear loss typically progresses to a mean level of 50–60 dB during the first 5–10 years (13). Seven of the eight patients also matched criteria for dVM and one for pVM. Thus, although these patients formally met MD criteria, VM with mild cochlear involvement appears to be the likely diagnosis.
Previous reports have already indicated an overlap of symptoms in VM and MD. Prevalence of migraine is increased in MD (1,14,15) and half of MD patients regularly experience migrainous symptoms with vertigo (1). In a study on six families with an index patient with MD and migraine, a familial clustering of migraine, episodic vertigo and MD was reported, indicating a genetic link between the two disorders (16). In a series of unselected MD patients, a history of migraine was associated with a higher incidence of bilateral cochlear involvement (56% in MD with migraine vs. 4% in MD without migraine), while the degree of hearing loss did not differ (17).
Substantial hearing loss has been regarded a discriminating feature of MD, while impairment of cochlear function in VM has been reported to be rare and only mild (6,18–20). However, all previous studies on VM have excluded patients with overt MD, regardless of concomitant migrainous symptoms, which leaves a potential overlap of more profound cochlear symptoms unaddressed. Our findings may be interpreted in two ways. A first possibility is that current diagnostic criteria of VM and MD are not sufficiently discriminative. A more compelling explanation, however, is a shared underlying pathomechanism (e.g. ion-channel dysfunction) in a subgroup of patients with migraine and vestibulo-cochlear symptoms meeting the criteria for VM or MD (17,21).
Other relevant differential diagnoses in our series included BPPV and psychogenic dizziness. One patient with typical BPPV at follow-up had been misclassified as pVM because of atypical positional nystagmus, which rarely occurs when multiple semicircular canals are involved (22,23). A second patient with dVM had concomitant BPPV. Given that the risk for BPPV is increased in patients with migraine (24), repeated nystagmus testing during the acute phase may be required to establish a final diagnosis (2,25). In two patients, non-vestibular psychogenic dizziness had become the prevailing symptom and possible underlying VM could not be distinguished with certainty anymore. Almost one fifth of VM patients had co-existing psychogenic dizziness at follow-up. An active depressive or panic disorder was detected in 16%. These findings are in line with previous studies reporting a high prevalence of psychiatric symptoms in patients with migraine (26,27) including patients with VM (28,29) and underline the importance to assess psychiatric co-morbidity in patients with VM.
Limitations of the study
This study, for the first time, assesses the validity of clinical criteria for VM in a large series of patients after a long follow-up period. The categories of dVM and pVM were explicit diagnostic criteria proposed by our group (10) and based on the ICHD (3), taking into account the specific features of vertigo in migraine. Assessment of patients included an extensive neurotological examination both at initial presentation and at follow-up to consistently rule out other causes of vertigo.
A limitation of this study is the lack of a biological marker for VM that could serve as diagnostic gold standard that unequivocally indicates the ‘true’ presence of VM in a given patient. Since such an indicator does not exist for most migraine syndromes including VM, the validity of our clinical diagnostic criteria was determined by referring to a follow-up diagnosis of dVM according to expert opinion as gold standard. This was done under the assumption that the additional clinical information obtained from the extended follow-up period enabled us to rule out other causes for vertigo such as MD more confidently and make the final diagnosis of dVM more robust compared with the initial diagnosis.
An additional limitation of our study is that validation of diagnostic criteria was performed as a single centre study. To extend the application of our criteria to a broader population of VM patients, our results should be reproduced in other dizziness centres.
Previous diagnostic criteria for VM
Criteria for VM in previous reports have been very heterogeneous. While some authors have restricted definition of VM to spontaneous rotational vertigo (9,30), others have included positional vertigo, imbalance or other perceptions of motion (4,5,8,10). Diagnosis of VM has been based on various combinations of clinical features, suggesting a migrainous origin of vertigo such as a lifetime or family history of migraine or a positive response to anti-migrainous treatment (4–6,9,10,19). The temporal correlation of vertigo and migrainous symptoms was not mandatory for diagnosis of VM in most surveys (4–6,8,9,19). However, taking together all epidemiological evidence, the mere association of vertigo and migraine in a patient is not generally sufficient to establish a diagnosis of VM with certainty. Both migraine and vertigo are highly prevalent in the general population with an estimated lifetime prevalence of 18% for migraine (31) and 7% for vertigo (32) and will frequently concur by chance alone.
Our criteria for definite VM therefore require the temporal association of vestibular and migrainous symptoms to increase diagnostic specificity. Although our patients were recruited through a specialized dizziness clinic and may not be representative of the general population, our results show that the majority of VM patients will meet stricter criteria of dVM as the disease progresses. Overall, 86% of VM patients fulfilled criteria for dVM at follow-up. This matches the findings of a previous study that determined the proportion of patients meeting criteria of dVM or pVM in 208 patients with benign recurrent vertigo. The majority of those with a history of migraine (n = 180) matched criteria for dVM (70%), while 30% met criteria for pVM (9). However, because a subgroup of patients may never experience vertigo and migrainous symptoms together, a provisional category of pVM should be maintained to allow greater sensitivity in the diagnostic tool.
Conclusion
Although vestibular symptoms are frequent in migraine, they are not well represented in the current classification of migraine and consensus diagnostic criteria are lacking. The proposed clinical criteria of dVM have a high validity for detection of patients with vertigo related to migraine when other relevant vestibular disorders are excluded. A provisional category of pVM should be maintained in a classification of VM to allow greater sensitivity. However, there remains an unresolved diagnostic overlap with MD in patients with VM and hearing loss. A common pathophysiological mechanism may be operative in a subgroup of patients with migraine and vestibulo-cochlear symptoms meeting diagnostic criteria for MD or VM.
Footnotes
Funding
This work has been supported by a grant ‘Rahel Hirsch’ by the Charité, Humboldt University Berlin.