Abstract
Background
Migraine with aura may mimic an acute ischemic stroke, so that an improper administration of thrombolytic treatment can expose migrainous patients to severe adverse effects.
Methods
This systematic review quantifies the relevance of migraine with aura among stroke mimics, checking for thrombolysis’ safety in these patients. We reviewed the literature after 1995, distinguishing from studies dealing with stroke mimics treated with systemic thrombolysis and those who were not treated with systemic thrombolysis.
Results
Migraine with aura is responsible for 1.79% (CI 95% 0.82–3.79%) of all the emergency Stroke Unit evaluations and it represents 12.24% (CI 95% 6.34–22.31%) of stroke mimics in the group not treated with systemic thrombolysis. 6.65% (CI 95% 4.32–9.78%) of systemic thrombolysis administrations are performed in patients without an acute ischemic stroke. Migraine with aura is responsible for 17.91% of these (CI 95% 13.29–23.71%). The reported rate of adverse events seems extremely low (0.01%).
Conclusion
Migraine with aura is the third most common stroke mimic, following seizures and psychiatric disorders; it is responsible for about 18% of all improper thrombolytic treatments. Despite the absence of strong supporting data, thrombolysis in migraine with aura seems to be a procedure with an extremely low risk of adverse events.
Introduction
Summary of studies about stroke mimics (SM) among patients evaluated in the emergency setting on the suspicion of an acute ischemic stroke. Migraine with aura (MA) is expressed as percentage with respect to all the SM in the study population.
DWI: Diffusion weighted imaging.
Summary of studies about stroke mimics (SM) among patients treated with rt-PA on the suspicion of an acute ischemic stroke. Migraine with aura (MA) is expressed as percentage with respect to all the SM in the study population.
AE: severe adverse effects after IVT administration; MCA: middle cerebral artery; MRI: magnetic resonance imaging.
Methods
A systematic review was carried out on PubMed using as keywords the following terms: “stroke mimic AND migraine”, “thrombolysis AND migraine”, and “stroke mimics AND migraine”. We limited our research to articles in English. The electronic bibliographic search was conducted in February 2017, with a final revision on 1 November 2017. All articles citing migraine as a stroke mimic (including case reports) published from 1995 to June 2017 were considered to be potentially eligible. Full texts were checked, and their references controlled carefully to ensure that all significant papers were included in the review. The search of the previously cited keywords yielded 93 scientific works overall. We chose not to take into account seven articles that were not published in English language (one in Dutch, one in German, one in Polish, two in French, two in Hungarian) and four articles published before 1995. Thirty five articles were included, while the other 47 works were carefully analyzed and considered not relevant, as they did not deal with qualitative or quantitative analyses of SM. For our statistical elaboration, we chose not to take into account three works about pediatric case series or transient ischemic attack mimics, in order to achieve a more homogeneous database. As a result, 14 articles dealt with the number of SM among all the SU evaluations, 18 works analyzed the number of SM treated with i.v. rt-PA.
Data on the prevalence of migraine with aura (MA) among SM were collected, as regards SU evaluations and IVT administered patients, separately. Adverse effects of thrombolysis in migrainous patients were reviewed. Using available data, analysis was performed using the random effect model, to determine the prevalence of migraine among SM and to evaluate the heterogeneity among articles. ANOVA Q test was used to search for a statistically significant difference between groups (IVT group vs. non-IVT group).
Literature review
Apart from the pioneering study by Norris and Hachinski reporting 13% of SM among 821 consecutive patients (19), the first systematic data goes back to 1995 when Dr. Rao’s American group analyzed 411 consecutive patients who presented to the emergency department with an initial diagnosis of stroke. Of these, 78 patients (19%) were SM, including postictal states, systemic infections, tumors, and toxic/metabolic disturbances (20). In the same year, the group in Cincinnati quantified at 4.3% (19 out of 446 emergency neurological evaluations) the burden of SM, with paresthesia or numbness of unknown causes in first place (three cases), followed by complicated migraine (two cases, 10.5% of the total), seizure, peripheral neuropathy and cranial nerve neuropathy (two cases each) (21). We can easily understand why a higher percentage of SM (9.5%, in 16 out of 169 patients) were found when general practitioners and hospital emergency service physicians (non-neurologists) performed the diagnosis of AIS (22). Ay and collaborators found that 27 out of 782 patients presenting with stroke-like deficits (3.5%) had no DWI alterations in brain regions clinically suspected to be responsible. Ten of the patients had an SM as migraine, seizure, functional disorder, transient global amnesia or a brain tumor (23). Cordonnier and collaborators reported a larger series of patients: Among 1402 patients admitted to the SU of the Lille University Hospital for a presumed neurovascular disorder (suspected cases with subarachnoid hemorrhage were excluded by the authors), 233 patients (16.6%) had no vascular disorder. MA was the diagnosis at discharge in 26 cases (11.2% of all the SM) and non-vascular acute headaches in another 25 patients (10.7% of SM) (24). Another retrospective study performed at the Michigan University considered 151 patients initially diagnosed with AIS and treated with rt-PA, without a Stroke Team evaluation; 10 of them (7%) had a different diagnosis at discharge, in particular, one patient had migraine with aura (NIHSS 3, 40 years) (25). Another report sets at 31% the burden of SM among emergency SU evaluations, with only three cases of migraine (26). The number found by Hemmen and collaborators is slightly lower: 25.3% of the Stroke Codes were SM (104 out of 411 patients), with five patients affected by MA; none of them received thrombolytic therapy (27). In the same year, the Dutch group of Vroomen studied 699 patients admitted to the Stroke Department: 15 people had migraine with prolonged aura, covering 47% of all the SM, which represented 4.8% of the total (32 cases); 13 of them were younger than 50 years (28). The proportion of acute stroke syndrome presentations that were not stroke was 22% according to the study by Tobin and collaborators: Migraine was one of the main causes (11% of all the SM) (29). Thanks to a Stroke Team evaluation, the Basel group reported only seven SM out of 250 patients treated with rt-PA (2.8%), without adverse effects except for minor gastrointestinal bleeding (30). Winkler et al. did not find migraine among SM, as Uchino et al. did not in a following retrospective study: Nine patients out of 254 (3.5%) were SM, and conversion disorder was the main diagnosis at discharge (31). The Houston group has confirmed the safety of rt-PA administration in SM in the first 3 hours of symptom onset: In their retrospective study, 69 out of 512 patients (14%) had an SM, mainly represented by seizure (38%), complicated migraine (37%) and conversion disorder (21%). They were treated with rtPA, but none of them had a systemic hemorrhage, intracerebral hemorrhage or angioedema (32). The following year, Chen et al. confirmed this data: In front of just 1.4% of SM, none of them presented asymptomatic or symptomatic cerebral hemorrhages, or angioneurotic oedema, after thrombolytic therapy. Migraine with visual aura was the diagnosis at discharge of one patient (NIHSS 6, 75 years) (33). The same safety profile emerges from the Finnish data of Artto and collaborators. Fourteen cases (1.4% of the total) were SM, included two patients with migraine with aura (NIHSS 6, 44 years; NIHSS 7, 56 years) (34). An American study in 2011 highlighted the safety of rt-PA administration in patients with SM, even within 3–4.5 hours of symptom onset, despite the small sample size (89 patients). A gingival bleeding was detected in one patient. In this study, nine SM were identified (10%), and complicated migraine was the most frequent diagnosis (five patients, 55.6% of all the SM) (35). Another retrospective study sets at 10.4% (56/539 patients) the number of SM among all the SU evaluations: Complicated migraine accounted for 19.6% of them (11 patients), but in all these cases IVT was not complicated with sICH or orolingual edema (36). A following Californian case series reported 30 SM out of 193 IVT-treated patients, including five cases of migraine, without any hemorrhagic complications (37). Again, we can easily expect that the percentage of SM may significantly rise up if the first evaluation is made by a non-neurologist physician or by the emergency medical system (32.6%) (38). Neurologists trained in acute stroke management can set down this percentage to more acceptable values, as shown in a following Spanish study. Among 621 patients treated with IVT, 15 were SM (2.4% of the total), including one case of MA (NIHSS 6, 45 years) and three cases of HaNDL syndrome (Headache and Neurological Deficits with cerebrospinal fluid Lymphocytosis: NIHSS 12, 25 years; NIHSS 5, 38 years; NIHSS 11, 51 years) (39). No ICH or other major complications of IVT were detected. SM represented 7% of the patients treated with IVT in a Swiss case series: 23 SM included epileptic seizures (n = 14), MA (n = 3, 13% of the SM) (NIHSS 2, 30 years; NIHSS 5, 27 years; NIHSS 3, 27 years) and conversion disorder (n = 3). None of them experienced ICH, major bleeding or death because of IVT (40). These data are in line with those reported by Forster and collaborators: 6.5% of 648 suspected AIS patients (42 cases) received a final diagnosis other than acute cerebrovascular event: Three of them were detected as MA, while in one case IVT was complicated by orolingual edema (41). To date, the largest study on the safety of thrombolytic treatment in SM reports two cases of sICH following IVT administration in patients, who were finally recognized as affected by an epileptic seizure. This multicenter study gathered 5518 patients, 100 of which were SM, mainly represented by epileptic seizure (n = 41), psychogenic disorder (n = 28) or migraine (n = 12) (42). Another study sets migraine at 11% of all the SM, representing 13% of the emergency stroke care pathway evaluations (we excluded seven cases represented by TIAs) (43). Coming to more recent articles, complicated migraine covers 20% of those 75 SM that were treated with IVT, as part of a series of 516 consecutive patients suspected for AIS. One patient with an SM (1.3%) suffered from an ICH with substantial neurological deterioration within 12 hours after rt-PA infusion and died during the third day of hospitalization. Two patients experienced minor extracranial hemorrhagic complications, such as gingival or nose bleeding (16). A Michigan case series reports 38 SM (11.8% of the total), with MA accounting for three cases, and two cases of major systemic hemorrhages (44). In the same year, the research group of St. Louis reported 20 SM among 120 patients who were treated with IVT: In this case series, conversion disorder was the main non-cerebrovascular diagnosis (n = 14), followed by MA (n = 2). None of the SM population had any hemorrhagic complications, but the authors observed that there was a higher prevalence of SM among patients treated with IVT with the drip-and-ship paradigm compared with those directly admitted to the hub hospitals (45). In the same year, the Californian group of Spokoyny and colleagues found 8.5% of SM among patients treated with IVT: One case of migraine with aura has been reported, without any hemorrhagic complications (46). More recently, Quenardelle and colleagues found 521 SM out of the 1361 patients prospectively enrolled in their stroke care pathway: The authors reported 90 cases of MA, two cluster headaches and five other defined headache types, together accounting for 18.6% of detected SM (47). The rate of SM seems comparable between patients who are evaluated via telestroke (where neurologists are connected to hospitals in areas without on-site neurological expertise) and those evaluated in an SU (48). SM cover about 10% of the English case series, reported by Sivakumaran et al., where migraine was the main final non-vascular diagnosis (28%) (49). Tables 1 and 2 summarize the main studies focused on the theme: Here, data do not include cases of subdural hematomas, subarachnoid hemorrhages, intracranial hemorrhages, TIAs and cortical venous thrombosis that were occasionally considered in the different case series.
A few studies dealt with the number of mimics among patients with clinically suspected TIA. Apart from the abnormally high values reported by an Australian case series (50), the rate of mimics ranges from 10 to 48.5% according to scientific literature (51–54); a recent study set at about 10% the number of TIA mimics, and MA represented more than 60% of cases (55). A Swiss case series showed a higher TIA mimic percentage (18%), with migraine representing almost a quarter of causes; furthermore, headache was found to accompany about 3% of all transient ischemic attacks, even if it was significantly more frequent in patients with TIA mimics than in those with true TIA (OR 3.71) (56).
Disorders that resemble stroke are quantitatively comparable in children: The group in Philadelphia prospectively reported 21% of SM (30 out of 143 patients) in the pediatric age; headache was a presenting symptom in 30% of SM, three of whom were diagnosed as migraine, while the other six cases were classified as not benign (57).
Results
Studies about SM in thrombolytic treatments go from 2003 to 2016, while the ones about SM among SU evaluations cover a larger time interval, from 1995 to 2016. In this latter group, migraine was responsible for 1.18% of all emergency SU evaluations (CI 95% 0.61–2.27%), with a high degree of heterogeneity (I2 89.85, p < 0.001). There is a growing prevalence of migraine along with the year of publication, ranging from 0.12% (20) to 6.61% (47); this tendency was statistically significant (p = 0.011, Figure 1). The same linear correlation was not statistically significant if we consider articles dealing with migraine attacks treated with rt-PA (p = 0.399, Figure 2). In order to verify the hypothesis of a raising prevalence of SM over time (possibly favored by the efforts towards a shorter door-to-needle time), we analyzed data regarding the non-IVT group and we found that the linear correlation between the number of SM and year of publication did not reach statistical significance (p = 0.246). As articles dealing with the prevalence of MA among SM are only available after 2002 for both groups (i.e. SU evaluations and IVT administrations), the comparison between groups has been performed to include only those articles that were published after 2002. In this way, MA showed a higher prevalence in the non-IVT group with respect to the IVT group (1.79% vs. 1.12%), despite not being significantly different (p = 0.348, Table 3). A high degree of heterogeneity, in terms of prevalence of migraine, was demonstrated among the articles (I2 86.53, p < 0.001 for the IVT group, I2 89.66, p < 0.001 for the non-IVT group). The role of MA as an SM is highlighted by another datum: The relative prevalence of MA among SM is 17.91% (with CI 95% 13.29–23.71%) if we consider the IVT group, 12.24% (with CI 95% 6.34–22.31%) for the non-IVT group. Again, this difference does not reach statistical significance (p = 0.275, Table 4). Finally, in order to quantify the relevance of SM in the daily neurologist emergency activity, we evaluated their prevalence among IVT administered patients and in the non-IVT group; in the latter group, analysis included works by Norris and Cramer as well (19,38). Among patients treated with IVT, the number of SM was significantly lower, as expected (6.65% with CI 95% 4.32–9.78%), with a high degree of heterogeneity (I2 98.03%, p < 0.001) with respect to the non-IVT group. In this case, the prevalence of SM was 12.56% (CI 95% 8.65–17.90%), again with a marked heterogeneity (I2 97.92%, p < 0.001). This difference reached statistical significance, with p = 0.020 (Table 5).
Relation between absolute prevalence of migraine with aura and year of publication of the article. Data refer to works dealing with all Stroke Unit evaluations (non-IVT group). Relation between absolute prevalence of migraine with aura and year of publication of the article. Data refer to works dealing with stroke mimics treated with thrombolysis (IVT group). Prevalence of migraine with aura among all emergency neurological evaluations. Data refer to articles published after 2002. Prevalence of migraine with aura with respect to the total amount of stroke mimics. Data refer to articles published after 2002. Prevalence of stroke mimics in the IVT and in the non-IVT groups (for the non-IVT group, works by Norris and Cramer were included).
As regards the safety of IVT administration in SM (Table 2), a total of only six patients experienced serious adverse effects (0.01% of 582 patients): two major systemic hemorrhages (44), one orolingual edema (41); two patients suffered from symptomatic systemic hemorrhages, while in one case an sICH occurred with a fatal outcome (16,42).
From a descriptive analysis of the final diagnosis reported in patients who were suspected to suffer from an AIS, treated as if an AIS was present (i.e. IVT), but in the end recognized as SM, MA represents the third main cause of improper treatment. It came after seizures (about 30% of patients) and conversion/psychiatric disorders (being responsible for more than a quarter of cases, data not shown).
Discussion
The revision of the literature highlights that MA is responsible for more than 1% of all emergency SU evaluations, and it causes more than 1.5% of all thrombolytic treatments (Figure 3). It is the third most common SM (after seizures and psychiatric/conversion disorders), and its misdiagnosis as an AIS is an actual and not negligible matter, as it may lead to improper rt-PA administration in patients suffering from non-ischemic-aetiology neurological symptoms. The tendency over years to decrease door-to-needle time might have contributed to an increase in the prevalence of SM in the IVT group (58). Actually, MA diagnosis in the emergency setting has gradually become more frequent in the last two decades (Figure 1), in front of a stabilisation of the rate of SM, demonstrating an improving recognition of this clinical entity. As expected, its prevalence is lower in the IVT group with respect to the non-IVT group, but it is not significantly different. On the contrary, MA prevalence among SM is higher in the IVT group (where it is responsible for almost 18% of improper thrombolytic treatments) with respect to the non-IVT group (about 12%). This data – together with the overall percentage of SM, which is significantly lower in patients treated with rt-PA with respect to the non-IVT patients (as expected) – demonstrate that MA maintains a pivotal role as a neurological mimicker. Its diagnosis remains a clinical challenge, for neurologists too. In front of a young patient with an acute focal neurological deficit (as an MA sufferer generally is), not to perform a potentially life- or disability-saving treatment is not an easy choice at all. Door-to-needle time or stroke severity does not necessarily help in distinguishing SM from AIS (34,48), even if NIHSS scores tend to be lower in SM (49) and a clinical onset with global aphasia without hemiparesis should raise the suspicion of an SM (30,39,40). Migrainous patients show a tendency to arrive quickly at the hospital, often within 1 hour of clinical onset (59). SM patients are often younger, more frequently women, with fewer cardiovascular risk factors, and they show psychiatric disorders as the most common comorbidity (28,35,45,47,49). However, the same features should be cautiously evaluated as possible hints for a non-vascular origin of the symptomatology: They may be responsible for the misdiagnosis of an AIS as well, in the case of the so-called stroke chameleons. Recently, a clinical algorithm has been proposed as a useful tool for the identification of SM (60).
Absolute prevalence of migraine with aura (%) among patients evaluated in an emergency setting for a suspected acute ischemic stroke (first group) and among patients treated with rt-PA (second group).
As regards the high heterogeneity between articles, the setting of the evaluations is crucial: Lower SM percentages can be found in those centres where (stroke) neurologists regularly evaluate patients, with multimodal CT or MRI scans. Higher values are associated with visits performed by non-neurologist specialists. At the same time, we must remember that a referral centre for thrombolysis may gather more hyper-acute patients than others, explaining a marked difference between hubs and spokes in SM prevalence (selection bias) (61).
According to scientific literature, the rate of adverse effect is extremely low, allowing us to consider IVT administration in a patient with SM, and in particular with MA, as an acceptably safe procedure, with an excellent functional outcome; a procedure not affecting the favourable natural history of SM itself (36,62). Even if literature does not report ICH after rt-PA administration in a patient with MA, there is a non-ignorable burden of potentially harmful consequence: The bleeding risk rises in the absence of a careful check of the main well-known contraindications. The MR imaging protocol, including perfusion weighted imaging (PWI), may help in recognizing MA as the cause of an acute neurological symptom, especially in the acute phase (63,64), while T2*-weighted gradient recalled echo (T2*-GRE) and susceptibility-weighted imaging (SWI) sequences may help in ruling out amyloid angiopathy (65).
This challenging diagnosis in the emergency setting is further complicated by the fact that migraine patients may experience cerebrovascular accidents themselves, which could be misinterpreted as attacks of MA (stroke chameleons (66)). Migraine patients have a higher risk of ischemic stroke, especially in women who take oral contraceptives and smoke cigarettes (67–69). Twenty percent of patients younger than 50 years suffering from a cerebrovascular event are migrainous (70), a percentage that rises to 29% in further younger patients (<35 years (71)). Furthermore, headache is reported in 13% of AIS with a negative brain CT scan in the acute phase (72), 27% of a large cohort of AIS (73), while it characterizes 28% of posterior circulation stroke according to a more recent article (74). Unfortunately, the available scientific works are not detailed enough to qualify the type of pain. The importance of this matter is confirmed by another last datum: There is a significantly lower rate of thrombolysis among young AIS patients (<50 years old) with migraine compared with AIS patients without a past medical history of primary headache (75). Given all these considerations and our data (6.65% of SM in the IVT group), the optimal and acceptable rate of SM set by Saver in 2010 at 1% of case series (76) seems far from being achieved, even in specialized settings.
In conclusion, along with a growing ability to correctly recognise a migraine attack in the last two decades, MA is one of the main concerns in the differential diagnosis of an AIS in the emergency setting. It is responsible for more than 1% of all SU evaluations, and it represents a great neurological mimicker: About 18% of all improper thrombolytic treatments are administered in patients with an acute migraine attack. A detailed clinical evaluation is necessary to achieve a correct diagnosis. When this is not possible, in front of persistent clinical doubt it is advisable to proceed with rt-PA administration, given the multiple evidences of its safety in scientific literature. Emergency brain MRI may surely reduce misdiagnosis, but at the dangerous price of a procrastination of a potentially life-saving therapy.
Footnotes
Clinical implications
More than 6% of rt-PA administrations are performed in patients without an acute ischemic stroke (stroke mimics).
Migraine with aura is the final diagnosis in more than 1% of patients evaluated in the emergency setting for the suspicion of acute ischemic stroke, and it is responsible for about 1.8% of thrombolytic treatments.
Diagnosis of migraine with aura is still a challenge: It is the third most common stroke mimic (following seizures and conversion/psychiatric disorders), being responsible for about 18% of all improper thrombolytic treatments.
Despite the absence of strong supporting data, thrombolysis in migraine with aura seems to be a procedure with a low risk of adverse events.
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.