ABSTRACT: Background. Recent data suggest that percutaneous closure of interatrial septal defect (IASD) is associated with a reduction in the intensity, frequency and duration of migraine headaches. In this study we review our own data to determine if we can reproduce the relationship between IASD closure in patients with a history of a central nervous system event (stroke or a transient ischemic attack [TIA]) and migraine headaches (HA). Methods. Fifty-eight consecutive patients with a history of unexplainable stroke or TIA with the exception of the presence of an IASD were included in this retrospective study. Multiple variables were collected including age, gender, history of smoking, hypertension, diabetes, hypercholesterolemia, ejection fraction, anticoagulant use pre- and postprocedure, shunt grade across the IASD pre- and postprocedure, defect size and right-sided filling pressures. All patients with a history of migraine HA answered the Migraine Disability Assessment Test (MIDAS), a standardized migraine questionnaire. Descriptive analysis was performed on all variables and compared among migraine and nonmigraine HA patients. Pre and post closure intensity and frequency of migraine HA were compared. Results. Of 58 patients, 14 (24.14%) had migraine HAs prior to percutaneous closure. There were no significant differences among the migraine and nonmigraine HA groups except that the migraine HA sufferers were younger (p = 0.016). One patient with migraine HA died on follow up from complications of cardiomyopathy. Only 5 (38.5%) of 13 patients reported still having migraine HA post closure of IASD. The frequency (41.6 ± 36.4 vs. 9.3 ± 24.8; p = 0.005) and intensity (8.0 ± 1.9 vs. 2.1 ± 3.2; p = 0.001) of the migraine HAs were markedly reduced post closure at 759 ± 545.6 days (range 89–1,433 days). There was no relationship between the shunt grade and the frequency or intensity of migraine HA. Conclusion. We conclude that IASD closure in patients with history of migraine HA and stroke or TIA have a marked improvement in the frequency and intensity of their HA. Migraine HA resolved in 61.54% patients post closure. Larger randomized studies are needed to confirm these findings, which could have significant implications for sufferers of migraine HA.

       Interatrial septal defects (IASD) have been associated with an increased incidence of cryptogenic strokes.^1–7 Percutaneous closure of these defects is now widely performed and the safety of this procedure has been established. It is unclear, however, how effective IASD closure is in preventing further recurrences of strokes or transient ischemic events (TIA) when compared to traditional treatment with anticoagulation. Observational data, however suggest that IASD closure in patients with a history of cryptogenic stroke reduces the frequency and intensity of migraine headaches (HA) in migraine sufferers.^8–12 In this study, we retrospectively reviewed our own experience with IASD closure and its relationship to the frequency and intensity of migraine HA in patients with a history of unexplained strokes or TIAs.

Methods

       Fifty-eight consecutive patients with a history of unexplainable stroke or TIA with the exception of the presence of an IASD were included in this retrospective study that was approved by the Institutional Review Board at our institution. Patients were included in this study if they had a clinically or neuroradiologically confirmed ischemic stroke with no other identifiable cause, with the exception of an IASD. Patients were ruled out for cerebrovascular disease, arrhythmias and coagulation disorders. All patients have a documented IASD by saline contrast injection and have undergone closure of this defect percutaneously using either the Amplatzer occluder (AGA Medical, Plymouth, Minnesota) (for atrial septal defect [ASD]) or the CardioSEAL occluder (NMT Medical, Inc., Boston, Massachusetts) (for patent foramen ovale [PFO]). Patients were routinely followed at 1 month, 6 months and then annually postprocedure. Data were collected from medical records and by follow-up phone interviews.
       Multiple variables were collected including age, gender, history of smoking, hypertension, diabetes, hypercholesterolemia, ejection fraction, anticoagulants use pre- and postprocedure, shunt grade across the IASD pre- and postprocedure, defect size, size of IASD, degree of shunt, presence of atrial septal aneurysm and right-sided filling pressures.
       Patients with a history of migraine HA answered the Migraine Disability Assessment Test (MIDAS), a standardized migraine questionnaire. Two questionnaires relating to the intensity and frequency of migraines HA were filled by each patient; one describing migraine HA within 3 months prior to the index procedure (retrospective), and one within 3 months prior to the follow-up interview (prospective). Patients were also asked about whether they had aura defined as the presence of visual or focal neurologic symptoms prior to the onset of their migraine HA. Descriptive analyses were performed on all variables and were compared among migraine and nonmigraine HA patients. Pre- and postclosure intensity and frequency of migraine HA were compared using the Wilcoxon paired test.

Results

       Demographic and procedural variables are described in Tables 1 and 2, respectively. The majority of patients had a PFO (93.1%) and were treated with the CardioSEAL occluder (NMT) (82.8%). Closure of the defect was successful, with 84.5% of patients having

Table 1

no shunt at the end of the procedure in contrast to severe shunt in 88% of patients preprocedure. Patients were relatively younger with a low frequency of diabetics and current smokers.

Table 2

       Of 58 patients, 14 (24.14%) had migraine HA prior to percutaneous closure. There were no significant differences in the demographic (Table 3), and procedural variables among the migraine and nonmigraine HA groups except that the migraine HA sufferers were younger (47.8 ± 13.7 vs. 58.2 ± 14.9; p = 0.016). One patient with migraine HA died on follow up from complications of cardiomyopathy. Only 5 (38.5%) of 13 patients reported still having migraine HA post IASD closure. The frequency (41.6 ± 36.4 vs. 9.3 ± 24.8; p = 0.005) and intensity (8.0 ± 1.9 vs. 2.1 ± 3.2; p = 0.001) of the migraine HAs were markedly reduced postclosure at 759 ± 545.6 days (range 89–1,433 days). There was no relationship between the shunt grade and the frequency or intensity of migraine HA. Of 13 patients with migraine HA, 11 (84.6%) had an aura. Of patients with aura, 3 (27.3%) partially improved, 1 (9.1%) had no change in migraine HA and 7 (63.6%) had complete resolution of their HA. Among the 2 patients with no aura, 1 had resolution of his HA and the other experienced partial improvement.

Discussion

       Migraine HA has been described in approximately 20–40% of patients with IASD and a history of TIA or stroke, double the expected frequency of migraine HA in the general population.^8–12 Several observational studies have shown a relationship between migraine HA and closure of the IASD. In a study by Schwerzmann et al,10 closure of PFO in patients with migraine HA and aura

Table 3

reduced the frequency of migraines by 54%, and in patients with migraine but no aura, by 62%. In addition, in a study by Azarbal and colleagues,¹¹ migraine HA resolved completely at 3 months post-PFO/ASD closure in 60% of patients with migraine HA, and in 75% of patients with migraine HA and aura.¹¹ Furthermore, a study by Reisman et al¹² showed that migraine HA resolved completely in 56% of patients who underwent IASD closure, and 80% of patients showed a reduction in the mean number of migraine episodes per month. Finally, the Migraine Intervention with STARFlex technology (MIST) trial was a prospective, randomized, double-blind study that evaluated the effect of PFO closure on the incidence of migraine HA. In this study, 147 patients with migraine HA and aura were enrolled and randomized to PFO closure using the STARFlex device versus a sham procedure. Although MIST did not meet its primary endpoint of reducing migraine HAs by 40% at 6 months, there was an approximate 37% reduction in migraine burden (number of HAs multiplied by the length in hours of HA) in the STARFlex arm compared to 17% in the sham arm, a statistically significant difference.¹⁶
       Our study supports these observations, showing a reduction of migraine HA by 61.5% over a long-term follow-up period of 759 ± 545.6 days (range 89–1,433 days). In our study, both patients with and without aura, improved following closure of the IASD. It is unclear at this time whether the presence of aura is a predictor of migraine HA resolution following PFO closure. Although this study is not randomized or sham-controlled, the persistence of the reduction in migraine HA over a long follow-up period lessens the chance that this is a placebo effect. Our findings are in concordance with those reported by Giardini et al,¹³ where long-term migraine HA improvement was seen in patients after 3-year follow up.
       In this study we included 4 ASD patients. Two had strokes and 2 had TIAs. One of 4 ASD patients had migraine HA (25%). This incidence is not different from the incidence of migraine HA in patients with PFO (24.1% or 13/54 patients). Also, both ASD and PFO closures have been reported to link to migraine HA resolution. A recent study published by Azarbal et al11 has shown that both defects are associated with migraine HA, and closure of either defect resulted in improvement or resolution of the HA. It should be noted that the pathophysiology by which resolution of HA with IASD closure occurs is largely speculative, and both PFO and ASD might share similar mechanisms.
       The pathophysiology of reduction in migraine HA after PFO closure is not clear at this time, but it is possible that it relates to the reduction of platelet activation and aggregation or humoral factors that escape degradation by the lungs. In a study by Sharifi et al14 an abrupt and severe migraine HA occurred following closure of PFO in 5 of 13 consecutive patients that resolved almost immediately following administration of 300 mg of clopidogrel. In addition, Wilmshurst et al¹⁵ showed that the combination of clopidogrel for 4 weeks and aspirin for 6 months is superior to aspirin alone for 6 months in preventing migraine with aura after percutaneous closure of IASD. Our patients were placed on aspirin and clopidogrel postprocedure indefinitely and for 3 months, respectively. Finally, spasm can play a significant role in the occurrence of migraine HA, therefore, closure of IASD might not have an influence on patients with spasm-induced migraine.
       Closure of IASD for migraine HA cannot be recommended at this time, and randomized, more conclusive studies are needed prior to having PFO closure as a therapy for migraine HA patients.

       Study limitations. This study is limited by its small number of patients and retrospective design. However, consecutive patients who met predefined inclusion criteria were all included. Also, all patients answered a standardized migraine HA questionnaire comparing their HA before and after their procedure.