Familial and Sporadic Hemiplegic Migraine

A rare form of migraine with aura in which the aura includes fully reversible motor weakness; caused in some families by heterozygous variants in CACNA1A, ATP1A2, SCN1A or PRRT2, with SCN2A as a recently associated gene and a majority of cases still lacking a monogenic explanation.

Literature and regulatory information updated: 10 October 2026.

0. Initial Reference Table

1. Introduction

1.1. Definition and position in headache classification

Hemiplegic migraine is a form of migraine with aura in which the aura includes fully reversible motor weakness accompanied by at least one other visual, sensory or speech and language aura symptom. ICHD-3, published in 2018 by the IHS Headache Classification Committee, distinguishes it from typical aura because of its characteristic genetic and pathophysiological differences and codes it as 1.2.3; whenever an aura includes motor weakness, the attack is always classified as 1.2.3, even when brainstem symptoms are also present [1]. Orphanet describes the condition as a rare variety of migraine with aura in which motor weakness forms part of the aura, with two main forms defined by family history [3]. The ICHD-3 note recalls that “plegic” means paralysis, but most attacks feature weakness (paresis) rather than complete paralysis [1].

The familial form (FHM, ICHD-3 1.2.3.1) requires at least one first-degree relative (father, mother, sibling or child) or second-degree relative to have had attacks meeting the criteria for hemiplegic migraine. When genetic testing identifies the cause, the subform is specified at the fifth digit: FHM1 with a variant in CACNA1A, encoding a calcium channel on chromosome 19; FHM2 with a variant in ATP1A2, encoding a Na+/K+-ATPase on chromosome 1; FHM3 with a variant in SCN1A, encoding a sodium channel on chromosome 2; and 1.2.3.1.4, “FHM, other loci”, when testing does not identify a variant in these three genes [1]. The sporadic form (SHM, ICHD-3 1.2.3.2) describes an individual without a first- or second-degree relative meeting the criteria; attacks have the same clinical characteristics, some apparently sporadic cases carry known variants, and occasionally a relative develops the disorder later, requiring reclassification as FHM [1]. GeneReviews uses “simplex hemiplegic migraine” for an individual with a heterozygous pathogenic variant in ATP1A2, CACNA1A, PRRT2 or SCN1A and no first- or second-degree relative meeting the criteria [4].

1.2. Historical overview and molecular milestones

The molecular history of hemiplegic migraine began through an overlap with another inherited disorder. While studying cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL), mapped to chromosome 19, Joutel and colleagues observed patients with recurrent migraine with aura and, in 1993, demonstrated genetic linkage in two large FHM families, with a maximum lod score above 8 for markers also linked to CADASIL; the two loci lay within an interval of approximately 30 centimorgans [19]. In 1996, Ophoff and colleagues identified the gene, then called CACNL1A4 and now CACNA1A, encoding the brain-specific α1 subunit of the P/Q-type calcium channel, spanning approximately 300 kb and 47 exons: in FHM they found four missense variants in conserved functional domains, one on two different haplotypes in unrelated families, and in episodic ataxia type 2 (EA2) two variants disrupting the reading frame, showing that the two disorders were allelic channelopathies [20].

In 2002, Thomsen and colleagues published the Danish population-based study that identified 147 patients from 44 families, characterised the order and duration of aura symptoms and proposed revised criteria partly adopted in the subsequent classification [10]. In 2003, De Fusco and colleagues associated FHM2, linked to 1q23, with ATP1A2 variants, with loss of function of one allele as the mechanism, the first association of a Na+/K+ pump subunit with a genetic disorder [21]. In 2005, Dichgans and colleagues mapped a third locus to 2q24 through genome-wide linkage in families without CACNA1A or ATP1A2 variants and identified the missense variant Gln1489Lys (historical designation, = NM_001165963.4:c.4465C>A p.(Gln1489Lys)) in SCN1A, present in three families, strengthening the molecular link between migraine and epilepsy [22].

The fourth gene emerged from another area of neurology. PRRT2 was known to cause paroxysmal kinesigenic dyskinesia (PKD) and self-limited infantile epilepsy when Dale and colleagues described, in 2012, a family with the c.649dupC variant and heterogeneous paroxysmal disorders including hemiplegic migraine [23]. That same year, Riant and colleagues sequenced PRRT2 in 101 index cases with HM beginning before age 20 and no variants in the three known genes, finding variants in four [24]. In 2022, the same French group analysed 860 probands and proposed PRRT2 as the fourth autosomal dominant gene for FHM and SHM [25]; GeneReviews included it in the July 2024 revision [4]. In 2026, Riant and colleagues associated heterozygous SCN2A missense variants with familial and sporadic HM [5].

The GeneReviews chapter on FHM was published in July 2001 and fully updated in 2004, 2009, 2015 and 2021, with the 2024 revision devoted to PRRT2 [4]. Pietrobon's 2026 review integrates genetics and pathophysiology and concludes that FHM is autosomal dominant and monogenic in some patients and probably polygenic or oligogenic in others [12].

1.3. Nomenclature, reclassification and errors to avoid

The term FHM is used inconsistently to describe families in which several forms of migraine coexist, because most individuals with hemiparetic attacks alternate these with more frequent migraine attacks without hemiparesis; some families also have strictly identical attacks among affected members [4]. Pure HM attacks and typical aura attacks may alternate in the same patient and coexist in the same family [12]. The former “basilar migraine” is now migraine with brainstem aura (ICHD-3 1.2.2), requiring at least two fully reversible brainstem symptoms and the absence of motor or retinal symptoms; when motor weakness is present, the code is 1.2.3 [1]. OMIM retains ATP1A2-associated “familial basilar migraine” within the FHM2 entry [8].

Two inaccuracies in the sources should not be reproduced. The criteria reproduced in GeneReviews describe motor weakness as “usually longer than 72 hours”; ICHD-3 states the opposite: motor symptoms generally last less than 72 hours, although weakness may persist for weeks in some patients, and the note to the migraine with aura criterion allows motor symptoms lasting up to 72 hours [1, 4]. The GeneReviews statement about “the three FHM-related genes” predates the inclusion of PRRT2 and is outdated [4]. Some authors use “FHM4” for PRRT2, but this does not correspond to an ICHD-3 subform, which classifies these patients under 1.2.3.1.4 [1, 2].

1.4. Scope

FHM and SHM are covered with their genetic bases (CACNA1A, ATP1A2, SCN1A, PRRT2 and SCN2A), together with patients without an identified monogenic cause. The other CACNA1A allelic disorders (EA2 and spinocerebellar ataxia type 6, SCA6), SCN1A epilepsies, including Dravet syndrome, the PRRT2 epileptic and dyskinetic spectrum and CADASIL, an important differential diagnosis of aura with focal deficit and leukoencephalopathy, form part of the differential diagnosis.

2. Molecular Genetics and Pathophysiology

2.1. Genetic architecture

HM displays marked genetic heterogeneity. The GeneReviews table attributes 10–15% of FHM to CACNA1A, approximately 10% to ATP1A2, 2–3% to PRRT2 and less than 1% to SCN1A, leaving 66–80% without an identified gene [4]. These proportions refer to families and patients tested in published series, with variable selection, rather than population fractions; cohort-specific detection rates are analysed in Section 5.4. The same table gives, for each gene, the proportion of known variants detectable by sequencing or corresponding to deletions and duplications, measuring the spectrum of reported variants rather than the clinical sensitivity of testing [4].

Source: GeneReviews (Table 1), with sequencing data derived from HGMD; ATP1A4 has been reported in only one family, without confirmation [4]. Functional effects follow Pietrobon's review [12].

ClinGen classifies the SCN1A–FHM relationship as having “moderate” validity, whereas SCN1A relationships with Dravet syndrome, genetic epilepsy with febrile seizures plus (GEFS+) and developmental and epileptic encephalopathies (DEE) are “definitive” [26]. For ATP1A2, ClinGen recognises a single definitive relationship covering the autosomal dominant spectrum from HM to DEE, and a second definitive autosomal recessive relationship with lethal polymicrogyria syndrome [8].

2.2. CACNA1A and FHM1

CACNA1A, at 19p13.13, encodes the pore-forming subunit of the voltage-gated calcium channel CaV2.1 (P/Q type), blocked by ω-agatoxin IVA and insensitive to dihydropyridines; its historical aliases include CACNL1A4, MHP1, FHM, EA2, SCA6, APCA and HPCA [27]. The channel is highly expressed in the cerebellum and controls neurotransmitter release at central synapses [12]. More than 40 missense variants have been associated with FHM1 and SHM1, all with gain of function: increased opening probability, activation at more negative voltages and reduced inhibition by G proteins [12]. In knock-in mice, CaV2.1 current increases in cortical pyramidal cells and some trigeminal ganglion neurones, but not in cortical inhibitory interneurones, disrupting the balance between excitatory and inhibitory transmission [12].

The same gene produces opposing phenotypes depending on the mechanism. EA2 variants are generally truncating or splicing variants or multiexon deletions, with loss of function and haploinsufficiency; ClinGen assigns CACNA1A a haploinsufficiency score of 3 and a triplosensitivity score of 0 [4, 27]. SCA6 results from a small CAG repeat expansion at the C-terminus, encoding a polyglutamine tract [27]. CACNA1A undergoes alternative splicing: exon 37 has two variants (37a and 37b), both with reported pathogenic variants, and an alternative site in exon 47 contains the polymorphic CAG repeats, which the long transcript places in frame [4]. The gene also encodes the transcription factor α1ACT, which translocates to the nucleus and increases expression of several neuronal genes; both isoforms contain the CAG repeat [9]. Expansion thresholds differ between sources: the FHM GeneReviews describes expansions above 21 repeats, whereas the SCA6 GeneReviews considers up to 18 normal, 19 of uncertain significance and 20–33 fully penetrant [4, 9]. CAG expansions have not been found in FHM families [4].

2.3. ATP1A2 and FHM2

ATP1A2, at 1q23.2, encodes the Na+/K+-ATPase α2 isoform, which exchanges three sodium ions for two potassium ions per cycle; in the brain, the sodium gradient it creates drives astrocytic glutamate uptake [8]. The α2 isoform is expressed in neurones during embryonic and neonatal life and almost exclusively in astrocytes in adults, where it is coupled to the glutamate transporters GLT-1 and GLAST and is essential for clearing potassium and glutamate from the synaptic cleft [12]. This distribution distinguishes it from ATP1A3, exclusively neuronal in the central nervous system, and suggests a different pathophysiology from ATP1A3-related alternating hemiplegia of childhood (AHC) [28].

The first variants identified, NM_000702.4:c.2291T>C p.(Leu764Pro) and NM_000702.4:c.2659T>C p.(Trp887Arg), cause loss of α2 subunit function and haploinsufficiency, which became the general rule for FHM-associated ATP1A2 variants [4, 21]. More than 90 missense variants have been described, with complete or partial loss of function [12]. Homozygous truncating variants cause lethal polymicrogyria syndrome, with fetal akinesia, respiratory failure, microcephaly, polymicrogyria and facial dysmorphism [8, 29].

2.4. SCN1A and FHM3

SCN1A, at 2q24.3, encodes the sodium channel NaV1.1, predominant in the axon initial segment of inhibitory interneurones [12]. More than ten missense variants have been associated with FHM3 and SHM3, with an overall gain-of-function effect through destabilisation of the inactivated state and increased persistent current, unlike the loss of function responsible for most SCN1A epilepsies, including Dravet syndrome [4, 12]. The founding FHM3 variant, Gln1489Lys, lies in the hinged-lid domain responsible for fast inactivation; the homologous mutation in SCN5A accelerated recovery from fast inactivation two- to fourfold [22]. The late sodium current inhibitor GS967 reversed gain of function in vitro and in FHM3 mouse interneurones [12].

2.5. PRRT2

PRRT2, at 16p11.2, encodes a protein that interacts with SNARE proteins and regulates synaptic exocytosis, possibly through SNAP25 [30]. Most variants, including c.649dupC, generate unstable messenger RNA or a rapidly degraded truncated protein, causing haploinsufficiency; c.649dupC is present in 70–80% of individuals affected across the PRRT2 spectrum [6]. Loss of function disrupts synaptic neurotransmission, removes negative modulation of NaV1.2 and NaV1.6 channels (with intrinsic hyperexcitability) and alters positive modulation of CaV2.1 channels at active zones [4, 12]. The vast majority of c.649dupC carriers do not have HM, leading Pietrobon to regard PRRT2 as insufficient on its own and possibly a major modifier in a polygenic mechanism; based on the frequency of likely pathogenic variants in a large cohort, Riant and colleagues argue that it is the fourth autosomal dominant HM gene and should be tested in all patients [12, 25]. This difference in interpretation does not alter diagnostic management but matters for counselling, because penetrance across the PRRT2 spectrum is 50–90% [6].

2.6. SCN2A, other genes and polygenic contribution

In 2026, Riant and colleagues combined linkage and exome sequencing in a four-generation family and screened SCN2A in six additional families and 594 probands with familial or sporadic HM without variants in known genes: heterozygous c.4438A>G p.(Lys1480Glu) cosegregated with FHM, c.769T>A p.(Phe257Ile) was found in a second family and c.3955C>G p.(Arg1319Gly) in a sporadic case, all absent from gnomAD [5]. The ten carriers had typical HM attacks from childhood; two children with p.(Phe257Ile) had self-limited infantile seizures in their first year, and no carrier had permanent cerebellar ataxia, intellectual disability or recurrent febrile coma [5]. Functional studies using automated patch-clamp showed altered voltage-dependent and kinetic properties and abnormal firing in a computational model, with likely effective gain of function of the NaV1.2 channel [5, 12]. This finding has been published in only one series, without ClinGen curation in the sources consulted.

Other genes have been reported in few patients, with insufficient evidence to establish causality: SLC1A3 (EAAT1/GLAST), with the loss-of-function variant T387P (published designation, = NM_004172.5:c.1159A>C p.(Thr387Pro)) in one patient and a truncating variant in another; and SLC4A4 (NBCe1 cotransporter), whose biallelic variants cause proximal renal tubular acidosis associated with migraine, including HM in a minority, with near-total loss of transport in migraine-associated variants and an approximately 40% reduction in the others [12]. An increased burden of common and rare variants in several genes contributes to familial aggregation in families without variants in the classic genes [12]. In 2026, an Australian exome study of 182 FHM cases compared with 500 UK Biobank controls identified 27 high-confidence CNVs exclusive to cases, including around ATP1A2 and within PRRT2, enriched in Rho GTPase signalling and RNA polymerase II transcription pathways [31]. The same group tested machine learning models in 182 cases and 1035 controls: single nucleotide variants alone had limited discrimination (area under the curve 0.626 with GLMNet and 0.533 with XGBoost), improved by integrating CNVs; these are research findings without clinical validation [32].

2.7. Cortical spreading depression and animal models

The electrophysiological correlate of aura is cortical spreading depression (CSD), described by Leão: a slow wave, at 2–6 mm/min, of neuronal and glial hyperexcitation followed by sustained depression of activity [4]. Before or during aura, regional cortical blood flow decreases, generally remaining above the ischaemic threshold, followed by hyperaemia after one to several hours [1]. The eight FHM mouse models assembled by Pietrobon share increased susceptibility to CSD [12]. In FHM1 and FHM2, the common mechanism is excess synaptic glutamate through increased release (CACNA1A) or reduced clearance (ATP1A2), activating NMDA receptors above threshold with weaker stimuli and facilitating CSD initiation; in FHM3, increased persistent current in inhibitory interneurones leads to extracellular potassium accumulation [4, 12]. The SCN1A p.(Leu263Val) (NM_001165963.4:c.787C>G) model displays spontaneous CSD [4, 33]. A child homozygous for SCN1A L263V experienced potentially fatal apnoeas responsive to GS967, whereas heterozygous relatives had pure FHM [12].

Source: Pietrobon 2026, Table 3 (the table gives “L236V” and the text “L263V”; the text designation, consistent with GeneReviews, is used) [4, 12].

The severe S218L model phenotype correlates with the very low activation threshold of the mutant channel, which opens near the resting potential; in S218L carriers, cerebral oedema often follows mild head trauma [12]. Regarding pain, in adult Cacna1a R192Q mice, basal and evoked release of calcitonin gene-related peptide (CGRP) from the dura mater is unchanged and CGRP-induced dural artery vasodilatation is reduced, making CGRP-dependent meningeal neurogenic inflammation unlikely as the headache mechanism in these models; potassium-evoked release is increased in the intact trigeminal ganglion [12]. These preclinical data cannot predict clinical responses to anti-CGRP antibodies, discussed in Section 7.5.

2.8. Variant spectrum and nomenclature

The CACNA1A MANE Select transcript is NM_001127222.2, whereas the classic literature and GeneReviews use NM_001127221.2 numbering (and sometimes cDNA X99897, which contains three additional nucleotides, AAG at 3623_3625, and therefore an additional glutamate residue) [4, 34]. In MANE, positions downstream of the exon differing by one codon shift by one residue: classic T666M is p.(Thr665Met) in MANE, a designation already used in the Finnish study [34, 35]. For ATP1A2 (NM_000702.4), SCN1A (NM_001165963.4), PRRT2 (NM_145239.3) and SCN2A (NM_001040142.2), the MANE transcripts match those used in the sources [34]. The following table lists variants of clinical or historical significance, normalised using VariantValidator for GRCh38 and with ClinVar classifications where available.

Sources: VariantValidator and ClinVar, accessed on 10 October 2026 [34, 36]; phenotypes from GeneReviews and original studies [4, 5, 22, 35, 37–40]. The Riant 2026 abstracts do not specify the SCN2A transcript; the c. and p. descriptions agree on the MANE transcript [34].

2.9. Genotype–phenotype correlations

Correlations are best described for CACNA1A and ATP1A2, with limited data; the following designations are historical, and the corresponding MANE transcript notation is given in the variant table. In CACNA1A, R192Q and V1456L were associated only with hemiplegic attacks in Italian families; V714A with frequent unconsciousness; T666M, I1810L, R583Q and D715E with HM and ataxia; T666M with more frequent HM, severe coma and nystagmus; R583Q with stupor, fever and progressive ataxia, with fewer attacks after acetazolamide; D715E with less frequent hemiplegic attacks (64% of carriers); and S218L with delayed cerebral oedema and fatal coma after mild head trauma [4]. Ducros and colleagues found that six of nine CACNA1A variants in a French cohort were associated with HM and cerebellar signs (83% of carriers of these six variants had nystagmus, ataxia or both) and three with pure HM [11]. In ATP1A2, G301R is associated with a severe phenotype featuring epileptic seizures, coma and fever, and D718N and P979L with epileptic seizures and intellectual disability [4]. In FHM3, elicited repetitive daily blindness is a specific comorbidity, and HM attacks and epileptic seizures do not overlap in time; febrile coma and cerebellar ataxia are more frequent in FHM1 than FHM2 and have not been described in FHM3 [12]. No genotype–phenotype correlations are established in PRRT2, and the same variant produces different phenotypes within a family [6].

Different variants in the same gene produce pure HM or severe atypical HM with chronic symptoms, and relatives carrying the same variant may differ greatly in severity, suggesting as yet unknown modifiers [12]. In a large Portuguese family identified through systematic screening for hereditary ataxias in Portugal, CACNA1A R583Q in exon 13 (p.(Arg582Gln) on MANE) cosegregated with HM with or without cerebellar signs and with permanent cerebellar ataxia without migraine, with a maximum lod score of 4.47; the authors concluded that a single missense variant produced SCA6 and FHM phenotypes [4, 37]. This “SCA6 phenotype” is progressive ataxia caused by a missense variant, distinct from classic SCA6 caused by CAG expansion.

2.10. Penetrance, expressivity and mosaicism

FHM penetrance is high, estimated at approximately 80% [4]. Across the PRRT2 spectrum it ranges from 50% to 90%, and carriers may be asymptomatic [6]. In the Ducros cohort, 89% of 117 CACNA1A variant carriers had HM attacks [11]. Expressivity varies in age at onset, severity, attack frequency and associated signs, and family history may appear negative because of non-recognition or reduced penetrance [4]. Germline or somatic and germline mosaicism in a parent is possible when the variant is not detected in parental leucocytes [4].

2.11. Allelic disorders

Sources: FHM (Table 2), SCA6 and PRRT2 GeneReviews, and GeneGuidance [4, 6, 8, 9, 27, 29]. GeneReviews also lists, as SCN1A associations, one family with autism and one case of Rasmussen encephalitis with the variant NM_001165963.4:c.4723C>T p.(Arg1575Cys), which has conflicting pathogenicity classifications in ClinVar [4, 36]. In EA2, up to 33% of individuals with 21 or more CAG repeats had sufficient episodes for diagnosis, and interictal ataxia indistinguishable from SCA6 may develop after years [9].

3. Clinical Spectrum and System-Specific Manifestations

3.1. The typical attack: aura and headache

An HM attack is a migraine attack with aura in which the aura includes motor weakness. Aura symptoms arise in the cortex or brainstem: visual disturbance is most common (scotoma, photopsia, zigzag fortification spectra and sometimes diplopia), followed by sensory loss with numbness or paraesthesia of the face or a limb and dysphasia, usually with right-sided hemiparesis [4]. Motor symptoms typically begin in the hand and spread to the arm and face over 20–30 minutes; sudden onset mimicking stroke is rare [2]. Weakness is unilateral and rarely bilateral, and aura symptoms progress and succeed one another in the usual visual, sensory, motor, aphasic and brainstem order [12]. Aura and headache generally last longer than in typical aura [12].

The Danish population-based study described 147 patients from 44 families: most had all four aura symptom types (visual, sensory, aphasic and motor) and all had at least two; visual and sensory aura lasted significantly longer than in typical aura, often meeting prolonged aura criteria; 69% had brainstem symptoms during attacks; and headache occurred in 99% of attacks, unlike typical aura, in which aura without headache is more frequent [10]. ICHD-3 summarises these findings by stating that FHM very frequently features brainstem symptoms and that headache almost always occurs [1]. Deficits may last hours to days and outlast the headache; weakness generally lasts less than 72 hours but may persist for weeks in some patients [1, 4].

Attack frequency ranges from one per day to fewer than five over a lifetime, averaging two to three per year, with attack-free intervals ranging from 2 to 37 years and a decline with age [4]. Pietrobon's review likewise reports a highly variable mean frequency of three attacks per year [12]. Triggers include those of common migraine (foods, odours, physical exertion, stress), together with cerebral angiography and mild head trauma [1, 4].

3.2. Severe attacks

Some patients have severe attacks with impaired consciousness ranging from drowsiness and confusion to deep coma, often with extensive motor deficit and fever, sometimes with epileptic seizures and reversible cerebral oedema; hemiplegia and impaired consciousness may persist for weeks [4, 12]. ICHD-3 lists impaired consciousness including coma, confusion, fever and CSF pleocytosis as rare manifestations [1]. In the French cohort of 117 CACNA1A variant carriers, one-third experienced severe attacks with coma, prolonged hemiplegia or both, all with complete recovery [11]. Intermittent confusion and psychosis have been described, including in a probable CACNA1A family with ataxia, and attention and memory loss after a severe attack may last weeks to months [4].

Mild head trauma is the most dangerous trigger. Kors and colleagues described delayed cerebral oedema and fatal coma following trivial head trauma, with a lucid interval, in three individuals carrying CACNA1A S218L: two from a family with “extreme” FHM and the previously asymptomatic daughter of a patient with sporadic HM; the variant was absent in 152 controls, no common founder was identified and neuropathology showed Purkinje cell loss with relative preservation of granule cells [42]. Permanent motor, sensory, language or visual deficits after an attack are very rare, and HM-associated cerebral infarction and death are sufficiently rare that their occurrence should prompt investigation for another disorder combining migraine and stroke [4].

3.3. Permanent and interictal manifestations

The most characteristic permanent manifestation is cerebellar involvement with gaze-evoked nystagmus, slowly progressive ataxia and cerebellar atrophy [12]. Three figures circulate in the literature and measure different things. ICHD-3 states that chronic progressive cerebellar ataxia independent of attacks occurs in approximately 50% of FHM families, without distinguishing the gene [1]. The classic classification cited by Ducros and colleagues and reproduced in the SCA6 GeneReviews divides FHM families into pure FHM (80%) and FHM with permanent cerebellar signs (20%), using all FHM families as the denominator [9, 11]. The FHM GeneReviews reports cerebellar signs in 40–50% of CACNA1A-related FHM families, restricting the denominator to families with a CACNA1A variant, and adds that within these families up to 60% of affected individuals have permanent cerebellar signs [4]. In the same French cohort, CACNA1A variants were present in 15 of 16 probands from families with HM and cerebellar signs, but only 4 of 12 families with pure HM [11]. The differences between figures therefore primarily reflect the denominator chosen and cohort selection.

Cognitive impairment and learning difficulties, early-onset epilepsy independent of HM attacks and movement disorders complete the interictal picture [4, 12]. In the Dutch Leiden cohort, intellectual disability and progressive ataxia were observed only in carriers of variants in the three classic genes [43]. Neurological prognosis is often benign in the pure FHM group, without interictal epileptic seizures, ataxia, nystagmus or cognitive impairment [4].

3.4. Epilepsy and paroxysmal movement disorders

Epileptic seizures are recognised in FHM variant carriers. In ATP1A2 families they occur during severe attacks with coma and encephalopathy; focal seizures during severe attacks were described in two individuals with sporadic CACNA1A-related FHM, one with severe intellectual disability, congenital ataxia and early cerebellar atrophy [4]. De novo CACNA1A variants are a common cause of epileptic encephalopathy, which may be the first presentation in individuals who later develop FHM [4]. FHM may be mistaken for epilepsy and treated unsuccessfully as such [1]. PRRT2 is a major cause of childhood seizures, as SeLIE (formerly “benign familial infantile convulsions”) and PKD/IC (formerly “infantile convulsions and choreoathetosis”), and PKD may occur in carriers with or without HM [4]. PKD begins between 1 and 18 years, with attacks of dystonia, chorea, ballism or athetosis usually lasting less than one minute, triggered by sudden movement without loss of consciousness [44]; attacks range from 100 per day to one per week, decrease with age and may remit in adulthood, and approximately 30% of patients with PKD have a history of SeLIE [6].

3.5. Gene-specific phenotype

In FHM1 (CACNA1A), carriers of an identified variant have earlier onset, more severe symptoms and more interictal signs, such as nystagmus, episodic or progressive ataxia, learning difficulties and epilepsy, than individuals with FHM without an identified variant [4]. Age at onset was 11.7 ± 8 years (range 1–51) in the French Ducros cohort [4, 11]. Febrile coma and cerebellar ataxia are more frequent in FHM1 than in other forms [12].

In FHM2 (ATP1A2), severe attacks with coma, encephalopathy and epileptic seizures have been described in several families, and certain variants are associated with epilepsy and intellectual disability [4]. ATP1A2 may also cause AHC, with recurrent prolonged hemiplegic episodes mimicking HM [4].

In FHM3 (SCN1A), HM attacks and epileptic seizures do not overlap in time, febrile coma and ataxia have not been described, and elicited repetitive daily blindness (elicited repetitive daily blindness, ERDB) is a specific comorbidity [12]. In 2026, Cestèle and colleagues described four new families with FHM3 and ERDB: episodes beginning between infancy and age 9, three to ten per day, lasting up to 30 seconds, associated with the novel variants NM_001165963.4:c.4027G>T p.(Ala1343Ser), c.5350G>T p.(Val1784Phe) and c.5006C>T p.(Ala1669Val), alongside the previously reported c.4467G>C p.(Gln1489His) and c.4495T>C p.(Phe1499Leu), with gain of function confirmed in GABAergic neurones [40].

In PRRT2-associated HM, mean age at onset is 12–17 years; migraine occurs during or after motor aura, accompanied by visual, sensory or speech symptoms; triggers include stress, anxiety, light and heat; frequency ranges from several attacks per week to one per month; duration may exceed 72 hours; and the interictal neurological examination and magnetic resonance imaging (MRI) are normal [6]. In the French Riant cohort (2022), of 49 PRRT2 variant carriers, 26 had pure HM, 16 HM with another manifestation (epilepsy in 8, learning difficulties in 5, hypersomnia in 4, abnormal movements in 3), 3 epilepsy without migraine, 2 PKD without migraine and 1 was asymptomatic [25]. In the three-generation Japanese family studied by Suzuki-Muromoto and the 17 patients from six published families, onset was predominantly in adolescence, complicated phenotypes were shared within each family and five patients experienced spontaneous remission [45]. HM is an infrequent manifestation of the PRRT2 spectrum, and migraine with or without aura is rare [6].

The ten SCN2A variant carriers described in 2026 had typical HM attacks from childhood, without permanent ataxia, intellectual disability or recurrent febrile coma [5].

3.6. Patients with and without an identified variant

Comparing patients with and without a variant helps interpret test results. In the Dutch cohort, 208 patients with familial (199) or sporadic (9) HM and a CACNA1A, ATP1A2 or SCN1A variant were compared with 73 patients without a variant (49 familial and 24 sporadic): carriers had earlier onset, more affected relatives and attacks more frequently triggered by mild head trauma, with extensive motor weakness, brainstem symptoms, confusion and cerebral oedema, whereas most patients without a variant had a mild phenotype resembling common migraine; exome sequencing of 47 patients without a variant (33 familial and 14 sporadic) identified no new genes [43]. In Finland, among 293 patients with HM studied by exome sequencing in a collection of migraine families, only three had variants in the classic genes, and HM, with more severe aura and headache than non-hemiplegic aura, did not form a distinct clinical group, being interpreted as the extreme of the migraine with aura continuum [35]. In Denmark, 105 SHM probands and 483 first-degree relatives showed that probands had no increased risk of migraine without aura but a greatly increased risk of typical aura, and relatives had increased risk of both, indicating aetiological heterogeneity in SHM [46].

3.7. Imaging, electroencephalography and perfusion

No pathognomonic neuroradiological findings exist. The only abnormality on conventional imaging is vermian cerebellar atrophy in some CACNA1A-related FHM families [4]. During severe and prolonged CACNA1A attacks, transient diffusion abnormalities consistent with cytotoxic oedema have been described, followed by hemispheric atrophy usually contralateral to the hemiparesis, and MR spectroscopy has shown cerebellar abnormalities [4]. During an attack in a 20-year-old woman, MR angiography, computed tomography (CT) angiography and perfusion studies showed vasodilatation and hyperperfusion dissociated from neuronal function [47].

A 2025 systematic review assembled 40 articles with 65 patients: during aura, electroencephalography (EEG) showed low-amplitude activity in the hemisphere contralateral to hemiparesis in 5 of 7 recordings (71.4%); during headache and post-headache phases, contralateral medium- or high-amplitude slowing occurred in 28 of 51 (54.9%) and 9 of 24 (37.5%); and interictal EEG was normal in 26 of 33 (78.8%) [48]. Data come from case reports and series, with the corresponding selection bias.

3.8. Natural history across the lifespan

The disorder generally begins in the first or second decade, and attack frequency tends to decline with age [3, 4]. Follow-up of a multigenerational FHM1 family with T666M (eight patients, mean follow-up 13 years, range 9 to 15) showed that ictal manifestations, which usually improve after adolescence, may recur in adulthood, and that permanent neurological signs (examination, MRI, EEG and neuropsychological assessment) remained stable [49]. Indelicato and Boesch's review describes the CACNA1A spectrum as age-dependent: paroxysmal manifestations (hemiplegic aura, episodic ataxia, epilepsy, non-epileptic paroxysmal movement disorders) and chronic manifestations, with severe neurodevelopmental phenotypes in early-onset cases and neurobehavioural disorders and chronic cerebellar ataxia in older children and adults [50]. In a Barcelona paediatric series of 32 children with episodic neurological disorders and mean onset at 1.3 years (12 with paroxysmal torticollis and 9 with paroxysmal tonic upgaze), pathogenic variants were found in 6, two in CACNA1A, and a positive result was associated with developmental delay and tonic upgaze [51]. Spontaneous remission has been described in PRRT2-related HM [45].

4. Clinical Recognition and Warning Signs

4.1. When to suspect the disorder

HM should be considered in anyone with recurrent attacks of reversible unilateral motor deficit developing gradually over tens of minutes, accompanied or preceded by visual, sensory or language symptoms and followed by migraine-type headache [1, 2]. Progression from one symptom to another, gradual spread of weakness from hand to arm and face, and positive phenomena (scintillations, tingling) distinguish aura from the sudden onset and negative phenomena of a vascular event [1, 2]. A common error is confusing sensory loss with weakness, and ICHD-3 recommends a prospective aura diary to document symptoms [1].

In children and adolescents, suspicion arises with hemiplegic attacks and headache, particularly with a family history, and should increase when associated signs are present: nystagmus, episodic or progressive ataxia, paroxysmal tonic upgaze, developmental delay, epilepsy or paroxysmal dyskinesia [4, 51]. In infants and young children with hemiplegic episodes, AHC is the main differential diagnosis (Section 6). In adults without a previous migraine diagnosis, or with migraine and atypical symptoms, other serious neurological disorders must be excluded before attributing symptoms to migraine, and stroke assessment takes priority [4, 15].

Febrile encephalopathy with coma and focal deficit without a demonstrated infectious cause, particularly in a familial setting; disproportionate coma or cerebral oedema after mild head trauma; and brief episodes of blindness repeated several times daily in a child from an HM family should also raise suspicion [1, 40, 42].

4.2. Warning signs and emergencies

A first attack with hemiparesis is a cerebrovascular emergency until proven otherwise. Even with normal imaging and gradually progressive aura, clinical overlap, imprecise family history, FHM rarity and the severity of untreated stroke justify age-appropriate stroke assessment at presentation; with a negative family history, stroke or another central nervous system disorder must be considered even more strongly [4]. The case in Section 10 illustrates the opposite dilemma: HM mimicking acute stroke and exposing the patient to unnecessary thrombolysis [39].

Signs of attack severity include impaired consciousness (from drowsiness to coma), fever, epileptic seizures, signs of intracranial hypertension and deficit persisting beyond the patient's usual duration [4, 12]. Cerebral oedema may develop late, with a lucid interval, after mild head trauma [42]. The patient organisation CACNA1A Foundation advises treating CACNA1A-related HM with the urgency of an epileptic seizure, with hospital observation and repeat imaging if necessary, and recommends oximetry monitoring in patients with respiratory depression during attacks [52]. Cerebral infarction or death during an attack should prompt investigation for another diagnosis [4].

4.3. Newborn screening

There is no newborn screening for HM. Presymptomatic identification occurs only through cascade testing in a family with a known variant (Section 8).

5. Diagnosis: Criteria, Tests and Algorithms

5.1. ICHD-3 criteria

Diagnosis is clinical and based on ICHD-3 criteria, adopted by GeneReviews as consensus criteria [1, 4]. HM first requires attacks to meet migraine with aura criteria and then requires the aura to include both components of criterion B.

Source: ICHD-3, accessed at ichd-3.org on 10 October 2026 [1]. Weakness duration is not a mandatory criterion; criterion B requires only fully reversible weakness with visual, sensory or language symptoms [1].

Migraine with aura typically features aura developing over 5–20 minutes and lasting less than 60 minutes, followed immediately or after less than one hour by a 4–72-hour headache with nausea or photophobia; headache may be absent [4]. In HM, aura lasts longer, and associated brainstem symptoms do not change the classification to 1.2.2 [1, 10]. The familial category depends on available information about relatives and may change over time [1].

5.2. Exclusion of secondary causes

Sporadic cases generally require neuroimaging and other investigations to exclude other causes, and lumbar puncture may be necessary to exclude headache with transient neurological deficits and CSF lymphocytosis (HaNDL, ICHD-3 7.8) [1]. Imaging, CSF and EEG primarily exclude other causes and do not confirm the diagnosis [8, 14]. In the acute phase of a first attack, assessment follows the stroke protocol; the CACNA1A Foundation identifies MRI as the preferred investigation and CT as the initial investigation when MRI is unavailable [52]. Conventional cerebral angiography is contraindicated in HM because it may precipitate a severe attack [4, 14]. MR angiography and non-invasive perfusion studies may show dissociation between a reduced macrovascular signal and microvascular hyperaemia, which does not correspond to an occluded arterial territory [39, 47].

5.3. Genetic testing: indications and strategy

Molecular diagnosis is established by identifying a heterozygous pathogenic or likely pathogenic variant in a known gene; under American College of Medical Genetics and Genomics and Association for Molecular Pathology (ACMG/AMP) standards, these two classes are clinically equivalent, and a variant of uncertain significance (VUS) neither establishes nor excludes the diagnosis [4]. Testing confirms but does not exclude HM [14]. Identifying a variant specifies the ICHD-3 subform, guides family counselling, explains associated manifestations and informs some therapeutic choices, such as carbamazepine in FHM3 and PRRT2 [1, 6, 40].

Yield is low in most adults with adult-onset HM without nystagmus, epileptic seizures or other neurological features, and several studies have shown that the classic genes are not a major cause of adult SHM [4]. It is higher with early onset and associated signs: in the French series of 25 patients with SHM beginning before age 16 and both parents tested, 23 had a variant, de novo in 19 (76%); of the 19 with a de novo variant, 5 had pure HM and 14 ataxia, epilepsy or intellectual disability; none had an SCN1A variant [7]. In the Dutch cohort, additional symptoms such as chronic ataxia and epilepsy increased the likelihood of finding a variant [14, 43].

When sequential single-gene testing is chosen, GeneReviews suggests sequencing CACNA1A first, followed by targeted deletion and duplication analysis, when nystagmus, ataxia, paroxysmal tonic upgaze or developmental delay is present, and sequencing ATP1A2 first in the presence of epilepsy [4]. In current practice, a multigene panel including CACNA1A, ATP1A2, SCN1A and PRRT2 and differential diagnosis genes (for example ATP1A3, SLC2A1 and NOTCH3) is the strategy most likely to identify the cause while limiting VUS; exome sequencing is the most widely used comprehensive genomic test [4]. The Australian and New Zealand cohort tested PRRT2, PNKD, SLC1A3, SLC2A1, SLC4A4, ATP1A3 and ATP1A4 in addition to the three classic genes, and most probands had no candidate variant [53]. SCN2A, associated in 2026, should be included in updated panels [5].

Dosage analysis is essential. Approximately 25% of reported PRRT2 variants are deletions, including whole-gene deletions [4, 6]; eight of 30 probands with a PRRT2 variant in the French cohort had a whole-gene deletion [25]. In a 2026 Australian study, multiplex ligation-dependent probe amplification (MLPA) of the four genes in 170 probands with suspected FHM and no point variants or small insertions and deletions identified CNVs, confirmed by quantitative PCR, in 15 (approximately 9%): CACNA1A exon duplications in 6 and deletions in 2; ATP1A2 deletions in 2 and one duplication; SCN1A deletions in 3 and one duplication; PRRT2 deletions in 5 (one within a 26-gene microdeletion confirmed by SNP array); three patients had CNVs in more than one gene [13].

5.4. Diagnostic yield by cohort

These proportions are yields in cohorts with differing selection, criteria and technologies, rather than estimates of clinical test sensitivity. Selection explains the range from 7% of Finnish families to 63% in a Dutch referral cohort [12]. Riant and colleagues state that known genes explain fewer than 20% of cases, and Zielke and colleagues that missense variants in the three classic genes explain fewer than 20% of suspected cases [5, 13]. Russell and Ducros estimated that at least one-quarter of large families and most sporadic cases lack variants in the three classic genes [54].

5.5. Test limitations and report interpretation

Panel, exome or genome sequencing does not size CAG repeats, so a normal result does not exclude SCA6; when progressive ataxia is present, targeted repeat analysis should be requested [27]. Sequencing without CNV analysis misses some deletions and duplications in the four genes [4, 13]. Deep intronic variants and complex rearrangements may escape the exon sequencing underlying panels and exomes, and parental mosaicism may not be detectable in leucocytes [4]. A negative test does not exclude HM and is the most frequent result, particularly in adults with a mild phenotype [14, 43].

The report should state the gene, versioned reference transcript (preferably MANE Select), DNA- and protein-level HGVS designation, zygosity and ACMG/AMP class [4, 34]. In CACNA1A, the numbering difference between MANE NM_001127222.2 and historical transcript NM_001127221.2 requires attention: classic T666M appears as p.(Thr665Met) on MANE, and transferring historical numbering to MANE may designate a different variant, as with I1810L (Section 2.8) [34, 35]. For a VUS, segregation studies in affected and unaffected relatives, parental testing to assess de novo occurrence and, where available, functional studies help reclassification [4]. Incomplete penetrance limits the value of an asymptomatic carrier relative as evidence against pathogenicity [4, 6].

5.6. Assessments following diagnosis

After diagnosis, GeneReviews recommends neurological examination with quantitative eye movement assessment in individuals with nystagmus or complaints of incoordination or imbalance; EEG and neuroimaging when epileptic seizures occur; neuroimaging for reduced responsiveness because of cerebral oedema risk; assessment for movement disorders in PRRT2-related HM; genetic counselling; and family support through community resources, social services and home nursing when necessary [4].

5.7. Practical algorithm

For a first attack with motor deficit, the first step is stroke assessment under the local protocol. Once a vascular cause has been excluded and reversibility confirmed, a history of similar attacks with aura and headache permits application of ICHD-3 criteria; cases without a family history require neuroimaging and sometimes lumbar puncture to exclude HaNDL and other causes [1, 4]. Genetic testing by panel with CNV analysis, or exome with CNV analysis, is particularly indicated with onset in childhood or adolescence, a family history, interictal neurological signs, severe attacks or epilepsy; in adults with late-onset HM and no other manifestations, yield is low and the decision should be shared [4, 7, 43]. A negative result retains the clinical diagnosis and therapeutic management.

5.8. Prenatal and preimplantation diagnosis

Once the familial pathogenic variant has been identified, prenatal diagnosis and preimplantation genetic testing (PGT) are technically possible; professional and family perspectives on their use in a disorder with incomplete penetrance and variable expressivity may differ (Section 8) [4].

6. Differential Diagnosis

6.1. Acquired causes of transient focal deficit

The most important differential diagnosis is ischaemic or haemorrhagic stroke and transient ischaemic attack (TIA). Todd's postictal paralysis and other non-genetic causes of transient hemiparesis complete the immediate differential [4]. ICHD-3 recalls that aura may rarely be mimicked by carotid dissection, arteriovenous malformation (AVM) and epileptic seizure [1]. Stroke usually begins suddenly with negative phenomena, whereas aura progresses over minutes, with positive symptoms and successive modalities; none of these features alone is sufficient to exclude stroke in a first attack [1, 4].

HaNDL (ICHD-3 7.8) features headache episodes with transient neurological deficits and CSF lymphocytosis; distinguishing it from sporadic HM may require lumbar puncture [1]. Pleocytosis also occurs rarely in FHM attacks, making differential diagnosis difficult during an isolated attack [1]. In the SHM case described by Hotz and colleagues, headache began days after the deficits, previous scintillating scotomata were present and CSF protein was normal, whereas CSF protein is elevated in more than 90% of HaNDL cases; the authors also discussed stroke-like episodes following radiotherapy (stroke-like migraine attacks after radiation therapy, SMART), excluded by the absence of typical cortical enhancement [55].

6.2. Inherited disorders with migraine and focal deficit

With a family history of hemiparetic attacks and migraine, an infarct on imaging suggests mitochondrial encephalomyopathy with lactic acidosis and stroke-like episodes (MELAS), CADASIL or thrombophilia, such as factor V Leiden [4].

AD, autosomal dominant. Sources: FHM (Table 3), ATP1A3, GLUT1 deficiency and SCA6 GeneReviews [4, 9, 28, 56]. In mitochondrial disorders, mid- and late-pregnancy losses are frequent and often unrecognised [4].

6.3. Alternating hemiplegia of childhood

AHC begins before 18 months with paroxysmal hemiplegic episodes affecting limbs or the whole body, lasting minutes to days and sometimes weeks, remitting only during sleep and possibly recurring after waking, followed by persistent neurological dysfunction; the 2021 Mikati criteria require two mandatory and three major criteria, or two major and three minor criteria [28]. Flunarizine has been the most commonly prescribed prophylaxis for AHC for over 20 years, and abrupt withdrawal has been associated with deterioration [28]. The ATP1A3 GeneReviews lists other genes featuring HM or ataxia in the differential diagnosis: ADCY5, ATP1A2 (FHM and AHC), CACNA1A (FHM, EA2 and SCA6), CLDN5 (AHC), SCN1A (seizures and FHM) and SCN2A (AHC) [28]. The PRRT2 GeneReviews includes AHC and FHM among the differential diagnoses for patients with “paroxysmal hemiplegia” [6].

6.4. Metabolic disorders and episodic ataxias

GLUT1 deficiency includes paroxysmal motor (involuntary movements, ataxia, weakness or paralysis, choreoathetosis, dystonia, alternating hemiplegia) and non-motor (migraine, recurrent headaches, cyclic vomiting) events, and HM has been described in this syndrome [56, 57]. In inherited metabolic disorders, coma with focal signs such as hemiplegia or hemianopia occurs in leucinosis (maple syrup urine disease), ornithine transcarbamylase deficiency, methylmalonic and propionic acidurias and phosphoglycerate kinase deficiency, and stroke-like episodes occur in urea cycle defects, organic acidurias, mitochondrial diseases (MELAS, POLG), homocystinurias, congenital disorders of glycosylation and Fabry disease, among others [57]. The reference text recommends questioning vague diagnoses such as “basilar migraine” when ketoacidosis, hyperlactataemia or hyperammonaemia is present, even moderately [57].

Episodic ataxias are channelopathies without a metabolic marker; attacks last minutes to hours, and recognition matters because they may respond to treatment [57, 58]. Interictal nystagmus (downbeat and gaze-evoked) points to CACNA1A, CACNB4 or SLC1A3, interictal myokymia to KCNA1, dystonic or choreic attacks to PRRT2, and areflexia with pes cavus, optic atrophy and deafness (CAPOS syndrome) to ATP1A3 [57]. EA2 is allelic to SCA6 and FHM [58]. Paroxysmal dystonia and dyskinesia feature intermittent episodes without interictal signs and, like episodic ataxias and some epilepsies, often result from variants in ion channel genes [59]. In SeLIE and PKD, family history is positive in almost 90% of cases and familial PRRT2 variants are present in 87% [60].

6.5. Epilepsy

Focal seizures with postictal paralysis and HM itself may be confused, and FHM may be treated unsuccessfully as epilepsy [1]. Epilepsy and HM frequently coexist in the same patient or family in FHM2 and PRRT2-related HM, and de novo CACNA1A variants cause epileptic encephalopathies [4]. Normal interictal EEG in approximately four-fifths of patients with HM and contralateral slowing without epileptiform activity during attacks help distinguish the two conditions [48].

7. Treatments and Risk Management Strategies

7.1. General principles and evidence base

There is no disease-modifying treatment or medicine authorised for HM. GeneReviews summarises that symptomatic support during attacks is the only available treatment and prophylaxis rests on isolated reports [4]. No medicine is authorised by the US Food and Drug Administration (FDA) or the European Medicines Agency (EMA) for any CACNA1A-related disorder, and all HM uses are off-label, except medicines with a general indication for migraine prophylaxis [27]. Patients with HM have systematically been excluded from randomised trials of acute and preventive migraine treatment, and treatment is empirical, based on case series, expert opinion and extrapolation from migraine with aura [2, 12, 61]. There are no randomised clinical trials in FHM [8].

Management rests on four pillars: symptomatic treatment of headache and associated symptoms with medicines without vasoconstrictor effects; recognition and hospital treatment of severe attacks; prophylaxis when attacks are frequent or severe; and prevention of avoidable triggers, particularly head trauma and conventional angiography [4, 14]. Patients should have a written attack management plan and diagnostic information to present in emergency departments, because the differential diagnosis with stroke recurs during every severe attack; the CACNA1A Foundation provides an emergency guide for this purpose [39, 62].

7.2. Acute headache treatment

Paracetamol and non-steroidal anti-inflammatory drugs (NSAIDs) are frequently the first choice [14]. In 2024, the IHS published global recommendations for acute migraine treatment with “optimal” and “essential” levels for settings with differing resources, applicable to HM headache with vasoconstrictors excluded [17]. In the emergency department, the 2025 American Headache Society (AHS) adult guideline recommends offering intravenous prochlorperazine and greater occipital nerve block (level A); considers intravenous dexketoprofen, ketorolac and metoclopramide, subcutaneous sumatriptan and supraorbital nerve block level B; intravenous chlorpromazine, dexamethasone and valproate level C; advises against intravenous hydromorphone (level A); and classifies ketamine as having insufficient evidence (level U) [63]. In HM, sumatriptan from this list is contraindicated by US prescribing information (Section 7.3).

Newer acute medicines have a vascular profile different from triptans. Lasmiditan, a 5-HT1F receptor agonist, has no contraindications in US prescribing information; it is indicated for acute migraine with or without aura in adults at 50, 100 or 200 mg, with a maximum of one dose in 24 hours and no driving for 8 hours afterwards because of central nervous system depression [64]. The IHS considers lasmiditan non-vasoconstrictive, with low cardiovascular risk, and suitable for prescribing with comorbid cardiovascular disease [17]. Small-molecule CGRP receptor antagonists (gepants), such as rimegepant, ubrogepant and zavegepant, have no direct vasoconstriction or labelled contraindication for stroke or infarction; the IHS recommends caution in small vessel disease, Raynaud's phenomenon and high vascular risk, and notes that the AHS and Taiwan Headache Society allow their use when triptans are contraindicated for cardiovascular reasons, whereas the German Headache Society advises caution [17]. Rimegepant is used acutely at 75 mg, maximum 75 mg in 24 hours, with safety beyond 18 doses in 30 days not established; intranasal zavegepant is given at 10 mg, maximum 10 mg in 24 hours [65, 66]. None of these prescribing documents mentions HM, and the sources consulted contain no published data on their specific use in HM; the CACNA1A Foundation states that gepants and ditans “may be considered in the future” [52].

7.3. Triptans and ergot derivatives: prescribing information, guidelines and observational evidence

US sumatriptan prescribing information (IMITREX®, December 2025 version) contraindicates the medicine in patients with a history of stroke or TIA and a “history of hemiplegic or basilar migraine because these patients are at a higher risk of stroke”; it also contraindicates use in ischaemic coronary artery disease or coronary vasospasm, Wolff–Parkinson–White syndrome, peripheral or ischaemic bowel vascular disease, uncontrolled hypertension, following an ergot derivative or another 5-HT1 agonist within the preceding 24 hours, with monoamine oxidase A inhibitors and in severe hepatic impairment [15]. The same document describes cerebral haemorrhage, subarachnoid haemorrhage and stroke with 5-HT1 agonists, some fatal, and notes that in several cases the cerebrovascular event appeared primary and the triptan was administered in the mistaken belief that symptoms represented migraine; it therefore recommends excluding other serious neurological conditions before treating headache in patients without a previous migraine diagnosis or with atypical symptoms [15]. Injectable dihydroergotamine prescribing information states that the medicine “should not be administered to patients with hemiplegic or basilar migraine” and also contraindicates use with potent CYP3A4 inhibitors, in ischaemic heart disease, uncontrolled hypertension, peripheral arterial disease, sepsis, following vascular surgery, in severe hepatic or renal impairment and within 24 hours of a triptan [16]. The generic prescribing information consulted for ergotamine and caffeine does not explicitly mention HM, but contraindicates use in vascular disease, pregnancy and with potent CYP3A4 inhibitors because of ergotism and ischaemia risk [67].

Guidelines follow prescribing information. The IHS states that existing guidelines and guidance documents recommend against triptans in hemiplegic migraine, basilar migraine, recent ischaemic stroke, ischaemic heart disease, vasospastic angina and uncontrolled hypertension, adding that the lack of evidence that triptans increase infarction or stroke risk should be recognised: a 2015 systematic review found no association, and a cohort of 63,575 migraine patients, including 13,664 triptan users, found no association with stroke, cardiovascular events or mortality, although in low-risk patients [17]. Dihydroergotamine causes prominent vasoconstriction, particularly extracranially, and is contraindicated in cardiovascular disease [17]. Russell considers triptans and ergotamine contraindicated in SHM and FHM on pathophysiological grounds [68]. The FHM GeneReviews recommends generally avoiding vasoconstrictors because of stroke risk, whereas the PRRT2 GeneReviews recommends using triptans and dihydroergotamine “with caution” because of ischaemic vascular event risk [4, 6]. Russell and Ducros describe vasoconstrictors, including triptans, as historically contraindicated but frequently used off-label to stop headache [54]. The Leiden group allows that, although controversial, triptans may be prescribed when ordinary analgesics provide insufficient headache relief [14].

Retrospective series recorded no ischaemic events. In Finland, 76 patients with familial or sporadic HM who had used triptans at least once rated response at 6.9 (standard deviation 3.1) and adverse effect severity at 4.9 (standard deviation 3.3) on a 0–10 scale; there was no ischaemic stroke or myocardial infarction, and one patient reported prolonged neurological symptoms after one rizatriptan dose, without abnormalities on several MRI scans [69]. In a retrospective review at four US headache centres, 67 patients with basilar migraine and 13 with HM received triptans (40 and 5) or dihydroergotamine (27 and 8), without stroke or myocardial infarction; adverse effects were gastrointestinal discomfort, rash, cervical dystonia, nightmares and flushing with triptans, and chest tightness, dystonic reaction, transient asymptomatic T-wave inversion and agitation with dihydroergotamine [61]. The authors themselves calculated that, with samples this small, event rates compatible with the data could reach 4.5% in basilar migraine and 23% in HM; the absence of events in a few patients does not demonstrate zero risk [61].

The practical consequence is that the labelled contraindication and IHS recommendation remain, and triptan or ergot use in HM contradicts prescribing information and requires an individually documented decision, informed consent and absence of other vascular risk factors; observational evidence is reassuring but involves fewer than a hundred exposed patients and cannot estimate risk [15, 17, 61, 69]. In any event, a triptan should never be used during a first attack without an established diagnosis because of the risk of treating stroke as migraine [15].

7.4. Treatment of prolonged aura and severe attacks

No acute treatment has demonstrated efficacy for severe, prolonged aura [14]. Treatments used rest on reports and small series. Intranasal ketamine at 25 mg in 11 patients with severe, disabling FHM aura reproducibly reduced deficit severity and duration in 5 and provided no benefit in 6 [70]; its effect is limited by dose-dependent psychotomimetic effects [12]. Intravenous verapamil has been used as an abortive treatment for FHM and SHM in case reports [71, 72]; in a 20-year-old woman it reproducibly resolved headache within 20 minutes without affecting hemiplegia [47]. Intravenous verapamil doses used are absent from the abstracts consulted. Verapamil and acetazolamide are also described as abortive treatments in Pietrobon's review [12]. Intravenous nimodipine was used in a severe paediatric ATP1A2-related HM episode [73], whereas Russell considers it contraindicated in HM [68]; comparative data have not resolved this disagreement.

Severe attacks with impaired consciousness, fever, epileptic seizures or cerebral oedema require admission. Seizures are treated with anticonvulsants, and GeneReviews recommends corticosteroids in children with cerebral oedema to reduce potentially fatal manifestations, without specifying a dose [4]. The CACNA1A Foundation treatment page, which declares itself provisional while a panel prepares a consensus and is not a guideline, proposes early treatment before imaging with verapamil, acetazolamide or ketamine as first line; methylprednisolone, hypertonic saline and mannitol for cerebral oedema; ibuprofen, ketorolac and ondansetron for pain and nausea; avoidance of ergotamine, dihydroergotamine and triptans; preference for MRI; repeat imaging because oedema may be delayed; hospital observation for at least one night; and a three- to five-day treatment course; no doses are specified [52]. In the prolonged SHM case described by Hotz and colleagues, with CSF pleocytosis, prednisolone 500 mg/day was administered for five days; this was management in an isolated report rather than a standardised dose [55].

7.5. Prophylaxis: when to start and which medicines to use

The Leiden group proposes considering prophylaxis for more than two attacks per month or severe attacks [14]. The 2024 IHS preventive recommendations set a general threshold of at least four headache days per month, impact on personal, social or professional life, ineffective optimised acute treatment or frequent acute medicine use, and comment that prevention may be considered at a lower threshold in migraine with aura, “especially hemiplegic migraine and migraine with prolonged aura” [74]. For frequent attacks, GeneReviews recommends an acetazolamide trial in CACNA1A-related FHM and trials of usual prophylactic agents (tricyclic antidepressants, beta-blockers, calcium channel blockers and anticonvulsants, including topiramate, lacosamide, levetiracetam and valproate), emphasising that the correlation between response and HM type is limited and response evidence is anecdotal [4].

The most commonly used HM medicines are verapamil, acetazolamide, flunarizine and lamotrigine, the latter considered particularly useful for aura; anti-CGRP antibodies have been added more recently [12]. The Leiden group lists, without a strict preference order, flunarizine, sodium valproate, lamotrigine, verapamil and acetazolamide, and, with less evidence, topiramate, candesartan and pizotifen; propranolol is more controversial, but adverse effect evidence is insufficient to contraindicate beta-blockers [14]. Russell and Ducros include lamotrigine and acetazolamide [54]. Acetazolamide and verapamil, usually ineffective in migraine without aura, may help HM, but findings are inconsistent and many patients experience adverse effects or failure of several treatment lines [2]. In a single-centre cohort of 41 patients with non-polyglutamine CACNA1A variants followed prospectively for up to 20 years, most required prophylaxis with standard and off-label medicines such as acetazolamide, flunarizine, 4-aminopyridine and topiramate [2]. Acetazolamide is suggested specifically for CACNA1A-related FHM rather than FHM2 [8]; attacks decreased after acetazolamide in the R583Q family [4].

Evidence for individual medicines is low-level. Flunarizine was assessed in a retrospective audit at a UK tertiary paediatric unit (1998–2009) involving 72 children and adolescents (median age 13 years), including 8 with SHM and 5 with FHM: the starting dose was 5 mg in most (2.5 mg in 3, 7.5 mg in 1 and 10 mg in 6), with a duration of 12 months; at least 50% attack reduction occurred in 57% (41/72) overall and in 85% of HM patients versus 51% of the others; adverse effects occurred in 15 (21%), with depression in 6, weight gain or increased appetite in 5, tiredness or sedation in 2 and headache worsening in 2, and discontinuation in 13 (18%) [75]. In the FHM1 family with the T666M variant followed for a mean of 13 years, interval treatment with non-selective calcium antagonists reduced attack burden and was well tolerated [49]. In an FHM2 family with the ATP1A2 variant NM_000702.4:c.2191A>G p.(Met731Val), sodium valproate monotherapy reduced attacks in one patient and combined valproate and lamotrigine stopped attacks in two [76].

Sodium channel blockers have a particular role in SCN1A- and PRRT2-associated forms. In PRRT2-related HM, GeneReviews recommends carbamazepine or oxcarbazepine on anecdotal evidence alone and reports that low-dose carbamazepine prevented attacks [6, 45]; in the Japanese series, low-dose carbamazepine was effective in three patients, and spontaneous remission in five supports considering treatment duration carefully [45]. The dose of up to 100 mg/day effective in most patients with PRRT2-related PKD refers to dyskinesia and should not be transferred to HM [6]. In the four FHM3 and ERDB families described in 2026, all four probands responded to low-dose carbamazepine titrated to response, without reported adverse effects: at 16 mg/kg/day, blindness decreased from 10 episodes daily to 1–2 weekly and migraines disappeared; at 7 mg/kg/day, blindness episodes fell from 3–4 to 1–2 daily and migraines became rare in a patient previously using a wheelchair because of hemiplegia; at 6 mg/kg/day, episodes fell from approximately 10 to approximately 1 daily [40]. Carbamazepine also showed efficacy in previous FHM3 with ERDB reports [12].

7.6. Reference doses of preventive medicines

None of the following medicines has an approved HM indication. Doses given are those in prescribing information or general migraine guidelines, or reported in the primary source when no other source exists, and are not validated HM doses.

The IHS reports that, following effective prevention with flunarizine or beta-blockers, only one-quarter of patients maintained response 18 months after discontinuation, and 75% worsened 1–28 months later (mean 7.2 months with flunarizine and 4.4 months with beta-blockers) [74]. Factors to consider include liver disease, bleeding dyscrasias, alcoholism, obesity and pregnancy with valproate, and depression and parkinsonism with flunarizine [74].

7.7. Antibodies and medicines targeting the CGRP pathway

Anti-CGRP monoclonal antibodies (eptinezumab, fremanezumab and galcanezumab targeting the ligand, and erenumab targeting the receptor) are recommended by the European Headache Federation (EHF) for episodic and chronic migraine and, by consensus, as first line in those requiring prevention; the 2019 guideline required previous failure of at least two classes [83]. The AHS also considers CGRP-targeted therapies a first-line option without requiring previous treatment failure [84]. The 2022 EHF guideline advises assessing efficacy after at least three months, considering a pause after 12–18 months, avoiding antibodies during pregnancy and breastfeeding and using them cautiously case by case in vascular disease, vascular risk factors and Raynaud's phenomenon; it does not mention aura or HM [83]. US prescribing information for antibodies and gepants includes warnings about new or worsened hypertension and Raynaud's phenomenon, and antibody information includes hypersensitivity and constipation with serious complications; none mentions HM [65, 66, 85–89].

HM evidence comprises reports and series assembled in a systematic review with individual patient data analysis published in 2026: of 432 records, six studies met criteria (four case reports and two series, 13 patients, 12 in quantitative analysis), from tertiary centres in Italy, Hungary and Japan, with very low certainty under GRADE (Grading of Recommendations Assessment, Development and Evaluation) [2]. From baseline to three months, monthly headache days fell from a median of 18 (interquartile range 15–25) to 9 (3–20), without statistical significance (p = 0.077); monthly aura days from 3 (2–8) to 0.6 (0–2.3) (p = 0.018); and Migraine Disability Assessment (MIDAS) score from 88 (66–154) to 28 (5–74) (p = 0.018). Ten of 12 patients (83.3%; 95% confidence interval [95% CI] 40.0–100.0) had a greater than 50% reduction in headache and aura days, and 9 of 12 (75.0%; 95% CI 34.3–100.0) a greater than 50% MIDAS reduction; no adverse effects were reported [2]. In the Japanese series, however, two patients had increased motor aura frequency and one developed new visual aura with galcanezumab, resolving after switching to fremanezumab [2, 90]. No patient received erenumab, preventing comparison of ligand and receptor inhibition [2].

Sources: Romozzi and colleagues' review (Table 1), Danno, Sottani and Hotz [2, 55, 90, 91].

In the Italian PRRT2 c.649dupC family, treatment criteria were genetically confirmed FHM, at least four headache days monthly and failure of at least two prophylactic treatments (amitriptyline with timolol; amitriptyline with lamotrigine; amitriptyline with flunarizine and acetazolamide); three women aged 54, 49 and 21 received standard-regimen galcanezumab with a 240 mg subcutaneous loading dose then 120 mg monthly; monthly headache days fell from 11.7 (standard deviation 2.08) to 1.67 at three months (p = 0.002) and six months (p = 0.009), and MIDAS from 80.3 to 6; aura frequency reduction, at least 50% in all, was not statistically significant [91]. Antibody doses follow prescribing information: erenumab 70 mg or 140 mg subcutaneously monthly; galcanezumab 240 mg loading then 120 mg monthly; fremanezumab 225 mg monthly or 675 mg quarterly, with a US extension to children and adolescents aged 6–17 weighing at least 45 kg with episodic migraine, at 225 mg monthly; eptinezumab 100 mg intravenously every three months, sometimes 300 mg [85–88].

Theoretical concern about haemodynamic effects of CGRP blockade in HM has so far had no clinical counterpart: evidence indicates no direct effect on intracranial circulation, and no significant cardiovascular events have been observed in trials and real-world practice in migraine generally [2]. In FHM1 mouse models, the meningeal CGRP pathway does not appear to contribute to pain (Section 2.7) [12]. The review authors acknowledge publication bias and lack of controls and advocate prospective registries [2].

7.8. Comparative regulatory status

Sources: Drugs@FDA, US prescribing information and EMA medicines database, accessed on 10 October 2026 [15, 16, 64–67, 77, 79, 81, 82, 85–89, 92, 93]. The European extension date for acute atogepant treatment and the content of fremanezumab efficacy supplements from 2020 and 2025 are absent from the sources consulted.

7.9. Treatment trials and studies

None of these studies is a randomised efficacy trial in HM. Retrospective series and case reports are subject to publication and selection bias, responses lack a comparison group, and spontaneous remission and natural age-related attack reduction may simulate efficacy [2, 4, 45]. Indirect comparison of response percentages across studies with different populations, outcomes and durations is unjustified. Relative reductions should be read alongside absolute rates: a reduction from 18 to 9 headache days monthly in 12 patients did not achieve statistical significance [2]. The two registered paediatric topiramate trials have not posted registry results, and the paediatric trial in topiramate prescribing information for migraine generally did not demonstrate efficacy [82, 94]. The N-acetyl-L-leucine trial measures ataxia rather than HM attacks [95]. A ClinicalTrials.gov search for “hemiplegic migraine” found only six studies, all old and completed: the two topiramate trials and four glyceryl trinitrate or CGRP provocation studies from the Danish Headache Center, without an active intervention trial specifically for HM [94].

7.10. Developments over the past twelve months

Between October 2025 and October 2026, developments can be classified by evidence level. In genetics, publications included the association of SCN2A with familial and sporadic HM (human study with functional data) [5], CNV detection in approximately 9% of probands negative for point variants (human study) [13] and CNV and machine learning studies without clinical validation (research) [31, 32]. Treatment publications included the first systematic individual patient data review of anti-CGRP antibodies in HM, with very low-certainty evidence [2], a prolonged SHM case without motor attacks over six months on fremanezumab [55] and the series of four FHM3 and ERDB families responding to carbamazepine [40]. Pietrobon's review (September 2026) proposes preclinical therapeutic hypotheses: memantine, a rapidly dissociating NMDA blocker used in Alzheimer's disease at 10–20 mg/day, reversed impulsivity in Atp1a2 G301R mice, prevented brainstem depolarisations and post-seizure mortality in homozygous Cacna1a S218L mice and reduced CSD in rats, and reduced HM attacks and seizures in a woman with very severe FHM2 (case report); negative allosteric modulators of GluN2B receptors, positive EAAT2 transporter modulators and sodium channel inhibitors preferentially targeting the inactivated state, such as GS967, remain preclinical [12]. Guideline developments included the American Academy of Neurology (AAN) and AHS pharmacological adult migraine prevention guideline, replacing the 2012 guideline and not mentioning HM in its abstract [96], and the AHS update on parenteral emergency treatment [63]. The phase 3 N-acetyl-L-leucine trial in CACNA1A disorders was registered in October 2025 [95]. A 2026 Japanese review and a 2025 review from the IHS iHEAD programme on migraine in monogenic disorders added no new recommendations [97, 98]. Searches of US prescribing information and the EMA database identified no HM-specific authorisations, applications or opinions during this period, which does not prove their absence.

7.11. Agents and circumstances to avoid

Vasoconstrictors should generally be avoided because of stroke risk, and conventional cerebral angiography is dangerous because it may precipitate a severe attack [4, 14]. Mild head trauma triggers attacks and, in S218L carriers, delayed cerebral oedema and fatal coma; the CACNA1A Foundation suggests helmets for patients whose attacks are triggered by mild trauma [42, 52]. Nimodipine is considered contraindicated by Russell, although used in a severe paediatric case [68, 73]. Valproate carries increased liver failure risk in POLG mitochondrial disease, relevant to the differential diagnosis with mitochondrial disorders [81]. Flunarizine is contraindicated in depression and parkinsonism [77].

7.12. Pregnancy and breastfeeding

The FHM GeneReviews has no specific pregnancy recommendations. General IHS recommendations advise strongly avoiding valproate and topiramate because of teratogenicity and avoiding candesartan and lisinopril; during breastfeeding, non-pharmacological treatment and peripheral nerve blocks are first line, followed by propranolol or amitriptyline [74]. Valproate is contraindicated for migraine prophylaxis during pregnancy and in women of childbearing potential without effective contraception [81]. The EHF advises avoiding anti-CGRP antibodies during pregnancy and breastfeeding [83]. No clinical pregnancy data exist for flunarizine, and its SmPC recommends caution [77]. For PRRT2-related HM treated with anticonvulsants, GeneReviews recommends discussing fetal risk, dependent on medicine, dose and gestational stage, before conception, and allows consideration of stopping treatment or switching to a lower-risk medicine in mild forms [6]. Improvement during pregnancy has been observed in PRRT2-related PKD [6].

7.13. Procedures, perioperative care and transition

Conventional angiography should be replaced by non-invasive techniques whenever possible [4, 14]. The sources consulted describe no specific anaesthetic or perioperative HM recommendations; in practice, vasoconstrictor avoidance and attention to head trauma and epileptic seizures apply. In children and adolescents, the IHS mentions beta-blockers and flunarizine at weight-adjusted doses with very limited evidence [74]; flunarizine is not indicated under 18 in the SmPC but has been used in paediatric series [75, 77]; topiramate has a US indication from age 12, without demonstrated efficacy at ages 6–16 [82]; and fremanezumab is the only anti-CGRP antibody with a US paediatric dose [87]. Transition to adult care should preserve the attack plan, genetic information and reproductive counselling and reassess prophylaxis, because attack frequency tends to decline but attacks may recur in adulthood [4, 49].

7.14. Summary of recommendations and disagreements

8. Reproductive Options and Family Counselling

8.1. Inheritance and recurrence risks

FHM and simplex HM caused by ATP1A2, CACNA1A, PRRT2 or SCN1A variants follow autosomal dominant inheritance [4]. Most people with FHM have an affected parent because diagnosis requires an affected relative; in simplex HM, the variant may have arisen de novo or been inherited from an asymptomatic parent because of incomplete penetrance [4, 12]. Studies of sporadic Dutch, French and Danish patients demonstrated de novo ATP1A2 and CACNA1A variants, frequent in early-onset SHM: 19 of 25 patients with onset before age 16 in the French series [4, 7]. The SCN2A variants in the second family and sporadic 2026 case illustrate both situations [5].

De novo occurrence is confirmed only with parentage testing; without it, the variant is “presumed de novo” [4]. An apparently negative family history may result from unrecognised disease in relatives or reduced penetrance and is confirmed only by parental clinical assessment and testing, including interview, neurological examination and targeted testing [4]. In PRRT2, the main phenotypes (HM, PKD, SeLIE and PKD/IC) vary among relatives carrying the same variant, and a child of a carrier with HM may present only with infantile convulsions or dyskinesia [4, 6].

8.2. Cascade and presymptomatic testing

Once the variant is identified, targeted testing is available for at-risk relatives; GeneReviews places this decision within genetic counselling and makes no specific presymptomatic testing recommendation [4]. ClinGen has not assigned clinical actionability to CACNA1A or ATP1A2; SCN1A “moderate” paediatric actionability refers to epilepsies rather than FHM3 [8, 26, 27]. Concrete benefits of knowing carrier status include avoiding conventional angiography and vasoconstrictors, attention to mild head trauma (particularly S218L), correct interpretation of a first emergency presentation and reproductive planning [4, 42]. In testing asymptomatic minors, these benefits should be weighed against incomplete penetrance and the absence of disease-preventive treatment.

8.3. Reproductive planning, prenatal and preimplantation diagnosis

The ideal time for counselling is before pregnancy, and it should be offered to affected or at-risk young adults [4]. Prenatal diagnosis and PGT are possible after identifying the familial pathogenic variant, and professional and family perspectives on their use may differ [4]. When molecular diagnosis is unconfirmed, DNA banking from an affected person permits future testing [4]. Counselling should also address preventive medicines with teratogenic risk, such as valproate and topiramate, and therapeutic plan review before conception (Section 7.12) [6, 74].

9. Special Considerations: Population Diversity and Epidemiology

9.1. Prevalence and what each figure measures

The only population-based study cited by GeneReviews and Orphanet is the Danish study by Thomsen and colleagues (2002), conducted in a population of 5.2 million through National Patient Register searches, review of over 27,000 headache and neurology outpatient records and public announcements: it identified 147 patients from 44 FHM families [10]. Based on this work, HM prevalence was estimated at 0.01%, with a male:female ratio of 1:3 and similar numbers of familial and sporadic cases [4]. Orphanet places point prevalence in the 1–5/10,000 category, with a mean of 10/100,000, for Europe (expert estimate) and Denmark [3]. This value refers to HM of all causes in a Nordic population; it does not measure each genetic form's prevalence and should not be extrapolated to other countries. FHM2, for example, corresponds to approximately one-tenth of this value by inference from the proportion of families rather than direct measurement [8]. ICHD-3 summarises that sporadic cases have approximately the same prevalence as familial cases [1].

Gene-specific proportions vary with population and selection (Section 5.4): 14% of population-based Danish families had a CACNA1A or ATP1A2 variant [38], and the Finnish cohort found three variants in 293 patients [35], whereas a Dutch referral cohort enriched for large families reached 63% [12]. In the French cohort of 697 consecutive probands, ATP1A2 was the most frequent gene among positive cases (42%), followed by CACNA1A (26%), PRRT2 (17%) and SCN1A (15%) [25].

9.2. Geography, founder populations and recurrent variants

Clinical and genetic HM research comes predominantly from Europe: France, the Netherlands, Finland, Italy, Denmark and Iceland, and the United Kingdom [4, 38]. CACNA1A T666M is the most common recurrent pathogenic variant, without a founder effect [4]. The S218L variants in Kors and colleagues' three patients likewise shared no common founder [42]. In Denmark, variant-positive families were large and multigenerational, and the six variants found (historical designations: CACNA1A R583Q, T666M and C1369Y, the latter = NM_001127222.2:c.4103G>A p.(Cys1368Tyr); ATP1A2 V138A, R202Q and R763C, = NM_000702.4:c.413T>C, c.605G>A and c.2287C>T) differed between families [38]. In Finland, ATP1A2 p.(Ala606Thr) and p.(Arg383His) and CACNA1A p.(Thr665Met) were found in few families [35]. Outside Europe, there are cohorts from Australia and New Zealand [13, 53], Japan [45, 90] and South Korea, the origin of the case in Section 10, with the same ATP1A2 p.(Ala606Thr) variant described in Finland [35, 39]. In the US, Mathew and colleagues' treatment series came from four headache centres [61].

In Portugal, the large family identified through systematic hereditary ataxia screening, with 15 examined patients and CACNA1A R583Q, showed coexistence of HM, HM with cerebellar signs and progressive ataxia without migraine within the same family [37]. Consulted sources describe no Brazilian HM data; this absence does not demonstrate that no publications exist.

9.3. Sex and special populations

The Danish study's male:female ratio of 1:3 reflects the female predominance in migraine [4]. In women of childbearing potential, preventive medicine selection is constrained by valproate and topiramate teratogenicity [74, 81]. In children, options with established paediatric doses are scarce (Section 7.13). In older people, flunarizine requires a lower dose and monitoring for parkinsonism and depression [77], and a first late-onset attack should be presumed stroke until proven otherwise [4].

9.4. Reference networks and patient organisations

ERN-RND, the European Reference Network for rare neurological diseases, includes cerebellar ataxia and hereditary spastic paraplegia groups and dystonia and paroxysmal disorder groups, providing case discussions through the CPMS platform and a registry; HM is not specifically named on consulted pages, and its likely place is within paroxysmal disorders and CACNA1A ataxias [18]. For forms with epilepsy (SCN1A, PRRT2, ATP1A2 with encephalopathy), the relevant network is ERN EpiCARE. The CACNA1A Foundation, a US parent-founded organisation, provides HM resources, an emergency guide, a contact registry, a natural history study and a physician directory [62]. GeneReviews also lists the National Headache Foundation and MedlinePlus FHM page, ATP1A2 and CACNA1A LOVD variant databases, the SCN1A variant database, HGMD and ClinVar [4]. The European Migraine & Headache Alliance (EMHA) federates European headache patient organisations.

10. Illustrative Clinical Case

In 2026, Lee, Kim and Cho described a 37-year-old Korean man from Busan who developed sudden severe headache and global aphasia and was found unresponsive approximately two hours after complaining of headache [39]. CT at the local hospital showed no abnormalities. He reached the referral centre approximately six and a half hours after last being seen well, beyond the 4.5-hour thrombolysis window, and was not thrombolysed; his National Institutes of Health Stroke Scale (NIHSS) score was 7 because of global aphasia, he was alert and had no definite focal motor deficit. Blood tests and CSF (normal pressure and composition; negative infectious pathogen sequencing, cultures and tuberculosis PCR) were normal, and EEG showed marked left hemispheric slowing without epileptiform activity [39].

MRI at seven hours showed normal diffusion, apparent diffusion coefficient and FLAIR sequences; time-of-flight MR angiography showed attenuated vascular signal in left middle and posterior cerebral artery territories without fixed stenosis, confirmed by contrast-enhanced MR angiography; post-contrast T1 showed less left parenchymal enhancement; and susceptibility-weighted imaging (SWI) showed prominent left cortical veins. The authors interpreted the findings as a “dissociated” pattern with reduced macrovascular signal and presumed microvascular hyperaemia (Figure 1, panels A–L, available at https://pmc.ncbi.nlm.nih.gov/articles/PMC13330094/figure/Fig1/; Figure 2, MR angiography at seven hours, at https://pmc.ncbi.nlm.nih.gov/articles/PMC13330094/figure/Fig2/) [39]. Aphasia resolved within 24 hours without specific treatment; right-sided numbness and headache persisted two further days; resolution was complete within 72 hours, and day-three SWI was normal [39].

Subsequent history revealed recurrent headaches with transient right-sided weakness since adolescence and a father with a similar history. Targeted genetic testing identified heterozygous ATP1A2 NM_000702.4:c.1816G>A p.(Ala606Thr), classified as pathogenic, establishing FHM2 (clinical and diagnostic timeline in Figure 3: https://pmc.ncbi.nlm.nih.gov/articles/PMC13330094/figure/Fig3/) [39]. The same variant had been described in Finnish families [35]. The authors emphasise HM as a stroke mimic, the need to avoid inappropriate thrombolysis and the value of integrating evolution, multimodal imaging and genetics; acknowledged limitations include lack of quantitative perfusion, CT angiography or perfusion and follow-up MR angiography [39].

The case illustrates three points. Personal and family history, once obtained, could have guided diagnosis from admission, but acute presentation requires stroke assessment, as GeneReviews recommends [4]. Perfusion or susceptibility imaging showing hyperaemia without a diffusion lesion supports aura and argues against arterial occlusion [39, 47]. The decision against thrombolysis reflected the time window rather than the HM diagnosis, established only later; the report does not define standardised management for patients with known HM presenting within the therapeutic window.

A second case published in 2026 illustrates a severe SHM attack: a 64-year-old Austrian man with non-Hodgkin lymphoma treated with whole-brain irradiation and stem cell transplantation, migraine with aura treated with sumatriptan and three to four annual episodes of aphasia and right hemiparesis, was admitted to a stroke unit with aphasia, right hemiplegia and reduced consciousness (NIHSS 21), without ischaemia or haemorrhage on MRI; CSF contained 146 cells/µL, he was intubated, received prednisolone 500 mg/day for five days and recovered from aphasia by day 14 and hemiparesis by day 21; exome sequencing identified no FHM-associated variant [55]. The authors discussed HaNDL and post-radiotherapy stroke-like episodes as differential diagnoses; following failure of lamotrigine titrated to 100 mg/day, fremanezumab 225 mg monthly was associated with six months without HM attacks [55]. Frequent previous sumatriptan use in this patient contradicts US prescribing information (Section 7.3), and high-dose corticosteroids reflect this report's management rather than a guideline recommendation [15, 55].

11. Prognosis and Long-Term Follow-Up

11.1. Prognosis

Neurological prognosis is often benign in pure FHM without interictal signs [4]. Attack frequency is low in most patients (mean two to three annually), declines with age and permits prolonged attack-free intervals, but attacks may recur in adulthood, as in the FHM1 family followed for a mean of 13 years [4, 49]. In the French CACNA1A carrier cohort, one-third experienced severe attacks with coma or prolonged hemiplegia, all recovering completely [11]. Permanent post-attack deficits are very rare and infarction and death exceptional, except with cerebral oedema following mild head trauma in S218L carriers [4, 42]. Long-term prognosis is primarily determined by permanent manifestations: slowly progressive FHM1 cerebellar ataxia, intellectual disability and epilepsy, concentrated in carriers of classic-gene variants and rare or absent in patients without a variant [12, 43]. In the family followed over a decade, permanent signs remained stable [49]. Spontaneous remission is possible in PRRT2-related HM [45]. Attention and memory loss after a severe attack may last weeks to months [4].

11.2. Surveillance

The written attack plan, including diagnosis, identified variant, contraindicated medicines and instruction to avoid conventional angiography, should accompany the patient and be reviewed at every visit [4, 62].

12. Summary and Final Considerations

Hemiplegic migraine is a rare form of migraine with aura in which the aura includes fully reversible motor weakness, with an estimated prevalence of 0.01% in Denmark and similar proportions of familial and sporadic cases. ICHD-3 criteria require migraine with aura and an aura including motor weakness and another visual, sensory or language symptom; weakness generally lasts less than 72 hours but may persist for weeks, and the GeneReviews transcription “longer than 72 hours” is an error. The distinction between familial and sporadic forms depends on first- or second-degree relatives' histories and may change over time.

The known genetic basis includes CACNA1A (FHM1, CaV2.1 channel gain of function), ATP1A2 (FHM2, astrocytic Na+/K+-ATPase loss of function), SCN1A (FHM3, inhibitory interneurone NaV1.1 gain of function), PRRT2 (loss of function, with interpretation still debated) and, since 2026, SCN2A. Animal models converge on increased susceptibility to cortical spreading depression. Most patients, particularly adults with mild phenotypes, have no identified variant, and HM in these cases appears to represent the extreme of a polygenic migraine with aura continuum. Genetic testing should use a panel or exome with CNV analysis and MANE transcripts, has higher yield with early onset and associated signs and does not exclude the diagnosis when negative.

The clinical spectrum ranges from brief hemiparetic attacks to severe attacks with coma, fever, pleocytosis, epileptic seizures and cerebral oedema, which may be delayed after mild head trauma. Permanent manifestations, particularly FHM1 cerebellar ataxia, intellectual disability and epilepsy, define prognosis; proportions of 20% of all FHM families, approximately 50% of families in ICHD-3 and 40–50% of CACNA1A variant families measure different things. A first attack requires stroke assessment, and HM must be distinguished from HaNDL, alternating hemiplegia of childhood, CADASIL, MELAS and epilepsy.

Treatment is empirical. Triptans and ergot derivatives are contraindicated in US prescribing information and discouraged by the IHS, although retrospective series have recorded no ischaemic events in small numbers of patients. Prophylaxis, considered at a low threshold, uses off-label flunarizine, verapamil, acetazolamide, lamotrigine, valproate and topiramate at doses from prescribing information or general guidelines; carbamazepine has a role in FHM3 and PRRT2-related HM; and anti-CGRP antibodies have favourable findings in small series with very low certainty. Conventional angiography and head trauma should be avoided. Genetic counselling addresses an autosomal dominant disorder with approximately 80% penetrance, frequent de novo variants in early-onset sporadic forms and residual germline mosaicism risk.

13. References

1. Headache Classification Committee of the International Headache Society (IHS). The International Classification of Headache Disorders, 3rd edition. Cephalalgia. 2018;38(1):1-211. doi:10.1177/0333102417738202. PMID: 29368949.
2. Romozzi M, Palermo M, Danno D, Katsuki M, Tosto F, Signorelli F, et al. Effectiveness and safety of anti-CGRP monoclonal antibodies in hemiplegic migraine: an individual patient quantitative analysis. J Headache Pain. 2026;27(1):56. doi:10.1186/s10194-026-02283-5. PMID: 41618146. PMC12930611.
3. Orphanet. Familial or sporadic hemiplegic migraine (ORPHA:569). Paris: INSERM; [data date 2026-06-23/2026-07-02; accessed 2026-10-10]. Available from: https://www.orpha.net/en/disease/detail/569
4. Jen JC. Familial Hemiplegic Migraine. 2001 Jul 17 [updated 2021 Apr 29; revised 2024 Jul 4]. In: Adam MP, Feldman J, Mirzaa GM, et al., editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–. Available from: https://www.ncbi.nlm.nih.gov/books/NBK1388/
5. Riant F, Thompson CH, Wafa SMA, Fenton TA, DeKeyser JM, Abramova TV, et al. SCN2A variants are associated with familial and sporadic hemiplegic migraine. Brain. 2026:awag109. doi:10.1093/brain/awag109. PMID: 41885191.
6. Yang K, Quiroz V, Ebrahimi-Fakhari D. PRRT2-Related Disorder. 2018 Jan 11 [updated 2024 Jul 4]. In: Adam MP, et al., editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–. Available from: https://www.ncbi.nlm.nih.gov/books/NBK475803/
7. Riant F, Ducros A, Ploton C, Barbance C, Depienne C, Tournier-Lasserve E. De novo mutations in ATP1A2 and CACNA1A are frequent in early-onset sporadic hemiplegic migraine. Neurology. 2010;75(11):967-972. doi:10.1212/wnl.0b013e3181f25e8f. PMID: 20837964.
8. GeneGuidance. ATP1A2 — gene record (clinical curation; version dated 26 Sep 2026).
9. Casey HL, Gomez CM. Spinocerebellar Ataxia Type 6. 1998 Oct 23 [updated 2019 Nov 21]. In: Adam MP, et al., editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–. Available from: https://www.ncbi.nlm.nih.gov/books/NBK1140/
10. Thomsen LL, Eriksen MK, Roemer SF, Andersen I, Olesen J, Russell MB. A population-based study of familial hemiplegic migraine suggests revised diagnostic criteria. Brain. 2002;125(Pt 6):1379-1391. doi:10.1093/brain/awf132. PMID: 12023326.
11. Ducros A, Denier C, Joutel A, Cecillon M, Lescoat C, Vahedi K, et al. The clinical spectrum of familial hemiplegic migraine associated with mutations in a neuronal calcium channel. N Engl J Med. 2001;345(1):17-24. doi:10.1056/nejm200107053450103. PMID: 11439943.
12. Pietrobon D. Hemiplegic migraine: genetics and pathophysiology. J Clin Invest. 2026;136(18):e208001. doi:10.1172/jci208001. PMID: 42741940. PMC13574153.
13. Zielke T, Sutherland HG, Maksemous N, Smith RA, Griffiths LR. Identification of Copy Number Variations in Familial Hemiplegic Migraine Genes in Suspected Hemiplegic Migraine Patients. Biomedicines. 2026;14(5):954. doi:10.3390/biomedicines14050954. PMID: 42193281. PMC13203776.
14. Pelzer N, Stam AH, Haan J, Ferrari MD, Terwindt GM. Familial and sporadic hemiplegic migraine: diagnosis and treatment. Curr Treat Options Neurol. 2013;15(1):13-27. doi:10.1007/s11940-012-0208-3. PMID: 23203776.
15. GlaxoSmithKline LLC. IMITREX (sumatriptan) tablets — prescribing information. NDA020132. DailyMed setid 584abe73-8290-4484-ff8e-5890831c095e, version 24, effective 2025-12-09. Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=584abe73-8290-4484-ff8e-5890831c095e
16. Gland Pharma Limited. Dihydroergotamine mesylate injection — prescribing information. ANDA215623. DailyMed setid 07a9112c-8c04-4849-8cb3-9463f58731d2, version 7, effective 2023-08-22. Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=07a9112c-8c04-4849-8cb3-9463f58731d2
17. Puledda F, Sacco S, Diener HC, Ashina M, Al-Khazali HM, Ashina S, et al. International Headache Society global practice recommendations for the acute pharmacological treatment of migraine. Cephalalgia. 2024;44(8):3331024241252666. doi:10.1177/03331024241252666. PMID: 39133176.
18. European Reference Network on Rare Neurological Diseases (ERN-RND) [Internet]. Tübingen: ERN-RND; [accessed 2026-10-10]. Available from: https://www.ern-rnd.eu/
19. Joutel A, Bousser MG, Biousse V, Labauge P, Chabriat H, Nibbio A, et al. A gene for familial hemiplegic migraine maps to chromosome 19. Nat Genet. 1993;5(1):40-45. doi:10.1038/ng0993-40. PMID: 8220421.
20. Ophoff RA, Terwindt GM, Vergouwe MN, van Eijk R, Oefner PJ, Hoffman SM, et al. Familial hemiplegic migraine and episodic ataxia type-2 are caused by mutations in the Ca2+ channel gene CACNL1A4. Cell. 1996;87(3):543-552. doi:10.1016/s0092-8674(00)81373-2. PMID: 8898206.
21. De Fusco M, Marconi R, Silvestri L, Atorino L, Rampoldi L, Morgante L, et al. Haploinsufficiency of ATP1A2 encoding the Na+/K+ pump alpha2 subunit associated with familial hemiplegic migraine type 2. Nat Genet. 2003;33(2):192-196. doi:10.1038/ng1081. PMID: 12539047.
22. Dichgans M, Freilinger T, Eckstein G, Babini E, Lorenz-Depiereux B, Biskup S, et al. Mutation in the neuronal voltage-gated sodium channel SCN1A in familial hemiplegic migraine. Lancet. 2005;366(9483):371-377. doi:10.1016/s0140-6736(05)66786-4. PMID: 16054936.
23. Dale RC, Gardiner A, Antony J, Houlden H. Familial PRRT2 mutation with heterogeneous paroxysmal disorders including paroxysmal torticollis and hemiplegic migraine. Dev Med Child Neurol. 2012;54(10):958-960. doi:10.1111/j.1469-8749.2012.04394.x. PMID: 22845787.
24. Riant F, Roze E, Barbance C, Méneret A, Guyant-Maréchal L, Lucas C, et al. PRRT2 mutations cause hemiplegic migraine. Neurology. 2012;79(21):2122-2124. doi:10.1212/wnl.0b013e3182752cb8. PMID: 23077016.
25. Riant F, Roos C, Roubertie A, Barbance C, Hadjadj J, Auvin S, et al. Hemiplegic Migraine Associated With PRRT2 Variations: A Clinical and Genetic Study. Neurology. 2022;98(1):e51-e61. doi:10.1212/wnl.0000000000012947. PMID: 34649875.
26. GeneGuidance. SCN1A — gene record (clinical curation; version dated 26 Sep 2026).
27. GeneGuidance. CACNA1A — gene record (clinical curation; version dated 26 Sep 2026).
28. Brashear A, Sweadner KJ, Haq I, Napoli E, Ozelius L. ATP1A3-Related Disorder. 2008 Feb 7 [updated 2024 Dec 5]. In: Adam MP, et al., editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–. Available from: https://www.ncbi.nlm.nih.gov/books/NBK1115/
29. Chatron N, Cabet S, Alix E, Buenerd A, Cox P, Guibaud L, et al. A novel lethal recognizable polymicrogyric syndrome caused by ATP1A2 homozygous truncating variants. Brain. 2019;142(11):3367-3374. doi:10.1093/brain/awz272. PMID: 31608932.
30. Inborn Metabolic Diseases: Diagnosis and Treatment. 7th ed. Saudubray JM, Baumgartner MR, García-Cazorla Á, Walter J, editors. Berlin: Springer; 2022. Chapter 30, section 30.6.1 (Table 30.4).
31. Alfayyadh MM, Zielke T, Maksemous N, Sutherland HG, Lea RA, Griffiths LR. Identification of case-specific copy number variants reveals novel genetic insights into familial hemiplegic migraine pathogenesis. Genomics. 2026;118(3):111245. doi:10.1016/j.ygeno.2026.111245. PMID: 41951059.
32. Alfayyadh MM, Maksemous N, Sutherland HG, Lea RA, Griffiths LR. Machine learning-based prediction of Familial Hemiplegic Migraine risk from genetic variants. Comput Biol Med. 2026;209:111699. doi:10.1016/j.compbiomed.2026.111699. PMID: 42033889.
33. Jansen NA, Dehghani A, Linssen MML, Breukel C, Tolner EA, van den Maagdenberg AMJM. First FHM3 mouse model shows spontaneous cortical spreading depolarizations. Ann Clin Transl Neurol. 2020;7(1):132-138. doi:10.1002/acn3.50971. PMID: 31880072.
34. Freeman PJ, Hart RK, Gretton LJ, Brookes AJ, Dalgleish R. VariantValidator: accurate validation, mapping, and formatting of sequence variation descriptions. Hum Mutat. 2018;39(1):61-68. doi:10.1002/humu.23348. PMID: 28967166. Available from: https://variantvalidator.org [accessed 2026-10-10].
35. Hiekkala ME, Vuola P, Artto V, Häppölä P, Häppölä E, Vepsäläinen S, et al. The contribution of CACNA1A, ATP1A2 and SCN1A mutations in hemiplegic migraine: A clinical and genetic study in Finnish migraine families. Cephalalgia. 2018;38(12):1849-1863. doi:10.1177/0333102418761041. PMID: 29486580.
36. National Center for Biotechnology Information. ClinVar [database]. Bethesda (MD): NLM; [accessed 2026-10-10]. Available from: https://www.ncbi.nlm.nih.gov/clinvar/
37. Alonso I, Barros J, Tuna A, Coelho J, Sequeiros J, Silveira I, et al. Phenotypes of spinocerebellar ataxia type 6 and familial hemiplegic migraine caused by a unique CACNA1A missense mutation in patients from a large family. Arch Neurol. 2003;60(4):610-614. doi:10.1001/archneur.60.4.610. PMID: 12707077.
38. Thomsen LL, Kirchmann M, Bjornsson A, Stefansson H, Jensen RM, Fasquel AC, et al. The genetic spectrum of a population-based sample of familial hemiplegic migraine. Brain. 2007;130(Pt 2):346-356. doi:10.1093/brain/awl334. PMID: 17142831.
39. Lee GH, Kim J, Cho JW. Dissociated neurovascular dynamics in ATP1A2-related familial hemiplegic migraine mimicking acute ischemic stroke. BMC Neurol. 2026;26(1):428. doi:10.1186/s12883-026-04930-5. PMID: 42115975. PMC13330094.
40. Cestèle S, Harper AJ, Marra S, Leroudier N, Rusconi R, Dhifallah S, et al. Elicited Repetitive Daily Blindness Associated With Gain-of-Function SCN1A Variants and Responsiveness to Sodium Channel Blockers. Neurol Genet. 2026;12(3):e200352. doi:10.1212/nxg.0000000000200352. PMID: 42040244. PMC13105197.
41. Jones KL, Jones MC, del Campo M. Malan syndrome. In: Smith's Recognizable Patterns of Human Malformation. 8th ed. Philadelphia: Elsevier; 2022. p. 206–209.
42. Kors EE, Terwindt GM, Vermeulen FL, Fitzsimons RB, Jardine PE, Heywood P, et al. Delayed cerebral edema and fatal coma after minor head trauma: role of the CACNA1A calcium channel subunit gene and relationship with familial hemiplegic migraine. Ann Neurol. 2001;49(6):753-760. doi:10.1002/ana.1031. PMID: 11409427.
43. Pelzer N, Haan J, Stam AH, Vijfhuizen LS, Koelewijn SC, Smagge A, et al. Clinical spectrum of hemiplegic migraine and chances of finding a pathogenic mutation. Neurology. 2018;90(7):e575-e582. doi:10.1212/wnl.0000000000004966. PMID: 29343472.
44. GeneGuidance. PRRT2 — gene record (clinical curation; version of 26 Sep 2026).
45. Suzuki-Muromoto S, Kosaki R, Kosaki K, Kubota M. Familial hemiplegic migraine with a PRRT2 mutation: Phenotypic variations and carbamazepine efficacy. Brain Dev. 2020;42(3):293-297. doi:10.1016/j.braindev.2019.12.007. PMID: 31902651.
46. Thomsen LL, Ostergaard E, Romer SF, Andersen I, Eriksen MK, Olesen J, et al. Sporadic hemiplegic migraine is an aetiologically heterogeneous disorder. Cephalalgia. 2003;23(9):921-928. doi:10.1046/j.1468-2982.2003.00614.x. PMID: 14616935.
47. Hsu DA, Stafstrom CE, Rowley HA, Kiff JE, Dulli DA. Hemiplegic migraine: hyperperfusion and abortive therapy with intravenous verapamil. Brain Dev. 2008;30(1):86-90. doi:10.1016/j.braindev.2007.05.013. PMID: 17614229.
48. Kedgan P, Anukoolwittaya P, Surawattanawong T, Thanprasertsuk S, Rattanawong W, Limotai C. EEG patterns in each phase of hemiplegic migraine: a systematic review and pooled prevalence meta-analysis. J Headache Pain. 2025;26(1):213. doi:10.1186/s10194-025-02150-9. PMID: 41087891. PMC12522619.
49. Indelicato E, Nachbauer W, Eigentler A, Donnemiller E, Wagner M, Unterberger I, et al. Ten years of follow-up in a large family with familial hemiplegic migraine type 1: Clinical course and implications for treatment. Cephalalgia. 2018;38(6):1167-1176. doi:10.1177/0333102417715229. PMID: 28856914.
50. Indelicato E, Boesch S. CACNA1A-Related Channelopathies: Clinical Manifestations and Treatment Options. Handb Exp Pharmacol. 2023;279:227-248. doi:10.1007/164_2022_625. PMID: 36592223.
51. Chinigioli M, Martí-Sanchez L, Yubero D, Xiol C, Olival J, Alcalá San Martín A, et al. Diagnostic value of genetic testing, with focus on CACNA1A, in children with episodic neurologic disorders: a single-centre retrospective study. Eur J Paediatr Neurol. 2026;60:50-57. doi:10.1016/j.ejpn.2025.11.005. PMID: 41370964.
52. CACNA1A Foundation. CACNA1A-Related Hemiplegic Treatment Options [Internet]. [accessed 2026-10-10]. Available from: https://www.cacna1a.org/
53. Sutherland HG, Maksemous N, Albury CL, Ibrahim O, Smith RA, Lea RA, et al. Comprehensive Exonic Sequencing of Hemiplegic Migraine-Related Genes in a Cohort of Suspected Probands Identifies Known and Potential Pathogenic Variants. Cells. 2020;9(11):E2368. doi:10.3390/cells9112368. PMID: 33126486. PMC7693486.
54. Russell MB, Ducros A. Sporadic and familial hemiplegic migraine: pathophysiological mechanisms, clinical characteristics, diagnosis, and management. Lancet Neurol. 2011;10(5):457-470. doi:10.1016/s1474-4422(11)70048-5. PMID: 21458376.
55. Hotz JF, Kaindl L, Krebs S, Vigl M, Ferrari J, Neumann C, et al. Reduced Frequency of Prolonged Sporadic Hemiplegic Migraine Attacks Following Fremanezumab Treatment-A Case Report. Eur J Neurol. 2026;33(2):e70514. doi:10.1111/ene.70514. PMID: 41589756. PMC12836861.
56. Wang D, Sands T, Tang M, Monani U, De Vivo D. Glucose Transporter Type 1 Deficiency Syndrome. 2002 Jul 30 [updated 2025 Mar 6]. In: Adam MP, et al., editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–. Available from: https://www.ncbi.nlm.nih.gov/books/NBK1430/
57. García-Cazorla Á, Saudubray JM. Clinical Approach to Inborn Errors of Metabolism in Paediatrics. In: Saudubray JM, Baumgartner MR, García-Cazorla Á, Walter J, editors. Inborn Metabolic Diseases: Diagnosis and Treatment. 7th ed. Berlin: Springer; 2022. Chapter 1, sections 1.4.1 (Table 1.6), 1.4.2 (Table 1.7), 1.5.2.6.
58. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 71.
59. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 139.
60. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 410–413.
61. Mathew PG, Krel R, Buddhdev B, Ansari H, Joshi SG, Spinner WD, et al. A retrospective analysis of triptan and dhe use for basilar and hemiplegic migraine. Headache. 2016;56(5):841-848. doi:10.1111/head.12804. PMID: 27062528.
62. CACNA1A Foundation [Internet]. [accessed 2026-10-10]. Available from: https://www.cacna1a.org/
63. Robblee J, Minen MT, Friedman BW, Cortel-LeBlanc MA, Cortel-LeBlanc A, Orr SL. 2025 guideline update to acute treatment of migraine for adults in the emergency department: The American Headache Society evidence assessment of parenteral pharmacotherapies. Headache. 2026;66(1):53-76. doi:10.1111/head.70016. PMID: 41321235.
64. Eli Lilly and Company. REYVOW (lasmiditan) tablets — prescribing information. NDA211280. DailyMed setid aea3358c-ff41-4490-9e6d-c7bf7b3de13f, version 16, effective 2026-01-08. Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=aea3358c-ff41-4490-9e6d-c7bf7b3de13f
65. Pfizer (distrib. Sportpharm LLC). NURTEC ODT (rimegepant) orally disintegrating tablets — prescribing information. NDA212728. DailyMed setid 86ef19d2-d6dc-4ac1-a793-dc549cc6bb9b, effective 2026-04-24. Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=86ef19d2-d6dc-4ac1-a793-dc549cc6bb9b
66. Pfizer. ZAVZPRET (zavegepant) nasal spray — prescribing information. NDA216386. DailyMed setid 1191475f-e999-49f4-a805-93ad08596b1b, version 4, effective 2025-12-26. Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=1191475f-e999-49f4-a805-93ad08596b1b
67. Trifluent Pharma LLC. Cafergot (ergotamine tartrate and caffeine) tablets — prescribing information. ANDA040590. DailyMed setid 4db7eece-2eef-bc1b-e063-6394a90a98cb, effective 2026-03-23. Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=4db7eece-2eef-bc1b-e063-6394a90a98cb
68. Russell MB. Management of sporadic and familial hemiplegic migraine. Expert Rev Neurother. 2010;10(3):381-387. doi:10.1586/ern.09.127. PMID: 20187861.
69. Artto V, Nissilä M, Wessman M, Palotie A, Färkkilä M, Kallela M. Treatment of hemiplegic migraine with triptans. Eur J Neurol. 2007;14(9):1053-1056. doi:10.1111/j.1468-1331.2007.01900.x. PMID: 17718700.
70. Kaube H, Herzog J, Käufer T, Dichgans M, Diener HC. Aura in some patients with familial hemiplegic migraine can be stopped by intranasal ketamine. Neurology. 2000;55(1):139-141. doi:10.1212/wnl.55.1.139. PMID: 10891926.
71. Yu W, Horowitz SH. Familial hemiplegic migraine and its abortive therapy with intravenous verapamil. Neurology. 2001;57(9):1732-1733. doi:10.1212/wnl.57.9.1732. PMID: 11706128.
72. Yu W, Horowitz SH. Treatment of sporadic hemiplegic migraine with calcium-channel blocker verapamil. Neurology. 2003;60(1):120-121. doi:10.1212/01.wnl.0000042051.16284.70. PMID: 12525732.
73. Dannenberg F, Prager C, Schmidt F, Tietze A, Bittigau P, Kaindl AM. Intravenous Nimodipine Treatment for Severe Episode of ATP1A2 Hemiplegic Migraine. Pediatr Neurol. 2020;112:71-72. doi:10.1016/j.pediatrneurol.2020.07.009. PMID: 32920306.
74. Puledda F, Sacco S, Diener HC, Ashina M, Al-Khazali HM, Ashina S, et al. International Headache Society Global Practice Recommendations for Preventive Pharmacological Treatment of Migraine. Cephalalgia. 2024;44(9):3331024241269735. doi:10.1177/03331024241269735. PMID: 39262214.
75. Peer Mohamed B, Goadsby PJ, Prabhakar P. Safety and efficacy of flunarizine in childhood migraine: 11 years' experience, with emphasis on its effect in hemiplegic migraine. Dev Med Child Neurol. 2012;54(3):274-277. doi:10.1111/j.1469-8749.2011.04154.x. PMID: 22268377.
76. Pelzer N, Stam AH, Carpay JA, Vries BD, van den Maagdenberg AM, Ferrari MD, et al. Familial hemiplegic migraine treated by sodium valproate and lamotrigine. Cephalalgia. 2014;34(9):708-711. doi:10.1177/0333102413520086. PMID: 24443394.
77. College ter Beoordeling van Geneesmiddelen (MEB). Samenvatting van de productkenmerken: Flunarizine CF 5 mg, capsules (RVG 56977). Revision 2021-07-28. Utrecht: CBG-MEB. Available from: https://www.geneesmiddeleninformatiebank.nl/
78. Bryant Ranch Prepack. Verapamil hydrochloride tablets — prescribing information. ANDA070994. DailyMed setid 006cf920-04ae-0b53-cb91-85481972abc2, effective 2024-05-22. Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=006cf920-04ae-0b53-cb91-85481972abc2
79. Heritage Pharmaceuticals (Avet). Acetazolamide tablets — prescribing information. ANDA205530. DailyMed setid 0700a910-79aa-4431-aecf-637fe14fcae8, version 11, effective 2026-05-12. Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=0700a910-79aa-4431-aecf-637fe14fcae8
80. GlaxoSmithKline LLC. LAMICTAL XR (lamotrigine) extended-release tablets — prescribing information. NDA022115. DailyMed setid 3e2c9a35-6a39-41d7-ad84-3c0bb8894b09, version 44, effective 2025-10-10. Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=3e2c9a35-6a39-41d7-ad84-3c0bb8894b09
81. AbbVie Inc. DEPAKOTE ER (divalproex sodium) extended-release tablets — prescribing information. NDA021168. DailyMed setid 0dc024ce-efc8-4690-7cb5-639c728fccac, effective 2026-03-31. Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=0dc024ce-efc8-4690-7cb5-639c728fccac
82. PD-Rx Pharmaceuticals. Topiramate tablets — prescribing information. ANDA090278. DailyMed setid 021005ab-6a00-40d8-9aee-20c91613b4d5, version 22, effective 2026-06-08. Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=021005ab-6a00-40d8-9aee-20c91613b4d5
83. Sacco S, Amin FM, Ashina M, Bendtsen L, Deligianni CI, Gil-Gouveia R, et al. European Headache Federation guideline on the use of monoclonal antibodies targeting the calcitonin gene related peptide pathway for migraine prevention - 2022 update. J Headache Pain. 2022;23(1):67. doi:10.1186/s10194-022-01431-x. PMID: 35690723. PMC9188162.
84. Charles AC, Digre KB, Goadsby PJ, Robbins MS, Hershey A, American Headache Society. Calcitonin gene-related peptide-targeting therapies are a first-line option for the prevention of migraine: An American Headache Society position statement update. Headache. 2024;64(4):333-341. doi:10.1111/head.14692. PMID: 38466028.
85. Amgen Inc. AIMOVIG (erenumab-aooe) injection — prescribing information. BLA761077. DailyMed setid b998ed05-94b0-47fd-b28f-cddd1e128fd8, version 25, effective 2026-07-14. Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=b998ed05-94b0-47fd-b28f-cddd1e128fd8
86. Eli Lilly and Company. EMGALITY (galcanezumab-gnlm) injection — prescribing information. BLA761063. DailyMed setid 33a147be-233a-40e8-a55e-e40936e28db0, version 17, effective 2026-06-05. Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=33a147be-233a-40e8-a55e-e40936e28db0
87. Teva Pharmaceuticals USA. AJOVY (fremanezumab-vfrm) injection — prescribing information. BLA761089. DailyMed setid 98e344ea-5916-4947-b6f2-4a76ccc04b6b, version 24, effective 2026-06-05. Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=98e344ea-5916-4947-b6f2-4a76ccc04b6b
88. Lundbeck Pharmaceuticals LLC. VYEPTI (eptinezumab-jjmr) injection — prescribing information. BLA761119. DailyMed setid 79065861-6aa5-4d1f-829f-3a6471286b36, version 32, effective 2026-06-05. Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=79065861-6aa5-4d1f-829f-3a6471286b36
89. AbbVie Inc. QULIPTA (atogepant) tablets — prescribing information. NDA215206. DailyMed setid 8c8ab8f4-32bd-497a-befa-70c8a51d8d52, version 19, effective 2025-09-30. Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=8c8ab8f4-32bd-497a-befa-70c8a51d8d52
90. Danno D, Ishizaki K, Kikui S, Takeshima T. Treatment of hemiplegic migraine with anti-calcitonin gene-related peptide monoclonal antibodies: A case series in a tertiary-care headache center. Headache. 2023;63(7):984-989. doi:10.1111/head.14591. PMID: 37366160.
91. Sottani C, Di Lazzaro G, Calabresi P, Pomponi MG, Tiziano FD, Bentivoglio AR, et al. Efficacy of galcanezumab in proline-rich transmembrane protein 2 (PRRT2)-associated familial hemiplegic migraine: A case series. Headache. 2025;65(2):377-381. doi:10.1111/head.14840. PMID: 39345003. PMC11794975.
92. U.S. Food and Drug Administration. Drugs@FDA: FDA-approved drugs [database]. Silver Spring (MD): FDA; [accessed 2026-10-10]. Available from: https://www.accessdata.fda.gov/scripts/cder/daf/
93. European Medicines Agency. Medicines for human use — EPAR (Aimovig, Emgality, Ajovy, Vyepti, Vydura, Aquipta, Rayvow) [database]. Amsterdam: EMA; [accessed 2026-10-10]. Available from: https://www.ema.europa.eu/en/medicines
94. ClinicalTrials.gov. Search: condition “hemiplegic migraine” [database]. Bethesda (MD): National Library of Medicine; [accessed 2026-10-10]. Available from: https://clinicaltrials.gov/search?cond=hemiplegic%20migraine
95. IntraBio Inc. Effects of N-Acetyl-L-Leucine on CACNA1A Disorders. ClinicalTrials.gov identifier: NCT07221292. First posted 2025-10-27 [accessed 2026-10-10]. Available from: https://clinicaltrials.gov/study/NCT07221292
96. Potrebic S, Tanveer S, Becker WJ, Burch R, Cooke LJ, Fenton T, et al. Pharmacologic Treatment for Migraine Prevention in Adults Practice Guideline Recommendations: Report of the AAN Guidelines Subcommittee and the American Headache Society. Neurology. 2026;107(7):e214881. doi:10.1212/wnl.0000000000214881. PMID: 42673560.
97. Shibata M, Saigoh K, Danno D, Kitamura E, Takizawa T, Kano Y. [Hemiplegic migraine: from historical perspectives to current management]. Rinsho Shinkeigaku. 2026;66(6):375-383. doi:10.5692/clinicalneurol.cn-002244. PMID: 42178246.
98. Mascarella D, Zhuang ZA, Gliga O, Santis F, Rosignoli C, Daputke A, et al. Migraine in monogenic disorders: Shedding light on new therapeutic targets. Cephalalgia. 2025;45(10):3331024251389241. doi:10.1177/03331024251389241. PMID: 41147997.
99. Rahbari R, Wuster A, Lindsay SJ, Hardwick RJ, Alexandrov LB, Turki SA, et al. Timing, rates and spectra of human germline mutation. Nat Genet. 2016;48(2):126-133. doi:10.1038/ng.3469. PMID: 26656846.