Clinical manifestations

Presentation and course

Reflex seizures are determined by the specific precipitating stimulus and the clinico-EEG response (Tables 1 and 2) (118; 82; 85).

Table 2. Reflex Seizures, Related Reflex Syndromes, and the Precipitating Stimuli

From (118); courtesy of the author and the publisher.

The precipitating stimulus. The stimulus evoking an epileptic seizure is specific for a given patient and may be extrinsic, intrinsic, or both. Extrinsic stimuli can be simple, such as flashes of light, elimination of visual fixation, and tactile stimuli, or they can be complex, such as reading or eating. The latency from the stimulus onset to the clinical or EEG response is typically short (1 to 3 s) with simple stimuli and long (usually many minutes) with complex stimuli. Intrinsic stimuli are elementary, such as movements, or elaborate, such as those involving higher brain function, emotions, and cognition (thinking, calculating, music, or decision-making).

Response to the stimulus. The response to the stimulus consists of clinical and EEG manifestations, alone or in combination. EEG activation may be subclinical only, that is without overt clinical manifestations. Conversely, ictal clinical manifestations may be triggered without conspicuous surface EEG changes.

Reflex seizures may be generalized, such as absences, myoclonic jerks, or generalized tonic-clonic seizures (GTCS), or they may be focal, such as visual, motor, or sensory. Reflex generalized seizures occur either independently or within the broad framework of certain epileptic syndromes. The same patient in response to the same specific stimulus may have absences, myoclonic jerks, and GTCS alone or in various combinations. Usually, absences and myoclonic jerks precede the occurrence of GTCS. Patients may have reflex and spontaneous seizures. Myoclonic jerks are by far the most common and manifest in the limbs and trunk or regionally, such as in the jaw muscles (reading epilepsy) or the eyelids (eyelid myoclonia with absences). GTCS may occur ab initio constituting the first clinical response or, more commonly, they follow a cluster of absences or myoclonic jerks. Secondary GTCS to focal, simple, or complex seizures are much less common than primary GTCS. Reflex absence seizures are common, constituting the response to a variety of specific stimuli, such as photic, pattern, fixation-off, proprioceptive, cognitive, emotional, or linguistic. It is recognized that absences are also common in self-induction. Reflex focal seizures are exclusively seen in certain types of reflex focal or lobular epilepsy, such as visual seizures of photosensitive lobe occipital epilepsy or complex focal temporal lobe seizures of musicogenic epilepsy.

The electroclinical events may be strictly limited to the stimulus-related receptive brain region only (such as photically induced EEG occipital spikes), spread to other cortical areas (such as in photosensitive focal seizures that propagate in extraoccipital areas), or become generalized (eg, in photoparoxysmal responses of idiopathic generalized epilepsy). Further, the electroclinical response to a specific stimulus may correspond to activation of regions other than those of the relevant receptive area, such as in primary reading epilepsy, which manifests with jaw myoclonic jerks. Conversely, reading may elicit electroclinical events strictly confined to the brain regions subserving reading, such as alexia associated with focal ictal EEG paroxysms. There is great variability in interindividual responses to the same stimulus.

A report reviewed the literature on extrinsic and intrinsic seizure triggers and discussed common mechanisms among them (75). Reflex seizures are typically associated with extrinsic sensory triggers; however, intrinsic cognitive or proprioceptive triggers have also been assessed. Emotional stress is the most frequent trigger in self-reported seizure triggers. The identification of a trigger underlying a seizure may be more difficult if it is intrinsic and complex, and if triggering mechanisms are multifactorial. The authors concluded that “since observability of triggers varies and triggers are also found in non-reflex seizures, the present concept of reflex seizures may be questioned.” They suggested “the possibility of a conceptual continuum between reflex and spontaneous seizures rather than a dichotomy” and discussed “evidence to the notion that to some extent most seizures might be triggered.”

The role of EEG is fundamental in establishing the precipitating stimulus in reflex epilepsies because it allows subclinical EEG, or minor clinical ictal events, to be reproduced on demand by application of the appropriate stimulus. However, there are cases in which the stimulus-seizure relationship is difficult to document, as in video game–induced seizures. Only 70% of these patients have EEG confirmation of photosensitivity with photoparoxysmal responses on intermittent photic stimulation; in the other 30%, seizures may be due to a single or a variety of other precipitating or facilitating stimuli. Sleep deprivation, mental concentration, fatigue, excitement, borderline threshold to photosensitivity, fixation-off sensitivity, proprioceptive stimuli (praxis), or more complex visual or auditory stimuli, alone or in combination, are all possibilities that are difficult to document objectively with EEG. There are also epileptic syndromes in which EEG “epileptogenic activity” is consistently elicited by a specific stimulus with no apparent clinical relevance. This is exemplified by certain cases of benign focal childhood seizures in which somatosensory (tapping) or visual (elimination of fixation and central vision) stimuli consistently elicit spike activity, though these children appear to have “unprovoked” seizures, which mainly occur during sleep.

The ictal manifestations of reflex seizures do not differ from those of spontaneous seizures of the same seizure type, for example, absence attacks or myoclonic seizures. Reflex seizures are, thus, often classified according to the triggering stimulus; this is reflected in an international classification proposal (34). Some typical clinical associations exist; for example, seizures induced by thinking are almost always myoclonic or clinically generalized attacks (Tables 3 and 4). It may also be clinically useful to think of precipitating stimuli for reflex seizures grouped according to their association with clinically generalized or bilateral EEG abnormalities or with focal EEG abnormalities, although this distinction does not constitute an element of the current proposed classification. This is also summarized in Tables 3 and 4.

Table 3. Reflex Seizure Types and Regional or Functional Triggers in Patients with Generalized Epileptiform Activity

Modified by the author after (169)

Table 4. Reflex Seizure Types and Regional or Functional Triggers in Patients With Underlying Focal Epilepsy

Visually induced seizures and epilepsies. Visually induced seizures and epilepsies are detailed in the MedLink clinical summaries “Visual-sensitive epilepsies,” “Idiopathic photosensitive occipital lobe epilepsy,” and “Eyelid myoclonia” (62; 118). Briefly, visually induced seizures and epilepsies are the commonest among reflex epilepsies. Visual seizures are triggered by the physical characteristics of the visual stimuli and not by their cognitive effects. Photosensitivity and pattern sensitivity are the 2 main categories, with frequent overlap.

Myoclonic jerks, GTCSs, and absences–in this order by prevalence–can occur in photosensitive patients. Some patients may have only 1 type, but most have any combination, particularly myoclonic jerks and GTCS.

Photically induced occipital seizures are much more frequent than originally appreciated after the use of intermittent photic stimulation in EEG testing. These may occur alone or progress to symptoms from other brain locations and GTCS. Extraoccipital focal seizures from the onset are exceptional.

Some recognized syndromes of idiopathic generalized epilepsy, such as juvenile myoclonic epilepsy, show a high incidence of clinical or, more frequently, EEG photosensitivity. This is much higher in Dravet syndrome (70%), Unverricht-Lundborg disease (90%), and other progressive myoclonic epilepsies.

“Fixation-off sensitivity” is a term coined by Panayiotopoulos to denote the form(s) of epilepsy and/or EEG abnormalities that are elicited by elimination of central vision and fixation.

Elimination of central vision and fixation is a specific precipitating stimulus, which, even in the presence of light, induces high-amplitude occipital or generalized paroxysmal discharges (115; 116; 117).

Seizures induced by reading and language. Seizures induced by reading are detailed in the article on primary reading epilepsy. Briefly, seizures induced by reading mainly manifest with jaw myoclonus (myoclonic variant), and less commonly with alexia (focal variant with alexia). Seizures usually evolve into generalized tonic-clonic seizures if reading persists. In the myoclonic variant, EEG discharges are brief and focal or bilateral synchronous, with a left-sided emphasis, whereas in the focal variant, they are prolonged and localized in the dominant posterior temporo-occipital junction. In the myoclonic variant, ictal functional neuroimaging shows multiple cortical hyperexcitable areas that are part of the neuronal network, which subserves the function of speech. In the focal variant, ictal functional neuroimaging shows mainly focal hyperexcitability in the dominant posterior temporo-occipital regions. Optimal treatment is with clonazepam for the myoclonic form and with carbamazepine (or other anti-epileptic drugs that are effective for focal seizures) for the focal variant with alexia.

Many patients with reading epilepsy have similar seizures elicited by other linguistic tasks (89). Thus, reading epilepsy may be considered as a variety of a more broadly defined reflex epilepsy involving a network for language (language-induced epilepsy) (89). Language-induced epilepsy involves seizure precipitation by speaking, reading, and writing (50). The seizures may be similar to those of primary reading epilepsy, and some patients report only a single effective trigger (eg, recitation alone) (64). Language-induced seizures may be confused with stuttering (103) and have been reported in families with idiopathic generalized epilepsy (153). Activation by reading and by other language tasks in the same patient can also occur in symptomatic epilepsies, as illustrated with video documentation by Canevini and colleagues (17). Language-induced epilepsy is not yet sufficiently defined to warrant classification as a separate syndrome independent of reading epilepsy. Reported cases are more heterogeneous than those of primary reading epilepsy.

Exceptionally, patients with autosomal dominant lateral temporal lobe epilepsy may have seizures precipitated by sound or speech (16). A 23-year-old patient had speech-induced seizures caused by an LGI1 mutation (C46R). He experienced short episodes of sensory aphasia in situations in which he was suddenly verbally addressed. Voices became distorted, and he could not comprehend despite hearing words. During a video-EEG recording, he was suddenly asked for the names of his siblings. He answered, but lost understanding of the further conversation and described how syllables floated together with an echoing character. With a versive movement to the right, another GTCS occurred. In the EEG, rhythmic 6-Hz activity built up in the frontotemporal areas starting on the left side with bilateral and posterior spreading. Postictal slowing was symmetrical, and no aphasia was noted on awakening.

Interestingly, an LGI1 mutation has also been reported in a patient with seizures triggered by answering the telephone (105). Telephone-induced seizures were reported in 3 patients who had onset during early adulthood of complex focal and secondarily generalized seizures exclusively triggered by answering the telephone (104). The seizures were stereotyped, with subjective auditory or vertiginous auras and inability to speak or understand the spoken voices. In 1 patient, a telephone-induced seizure arising from the dominant temporal lobe was recorded by means of video-EEG technique. In the interictal EEGs, temporal abnormalities were detected in all cases. The patients had normal neurologic examinations and normal brain imaging.

Seizures induced by simple and complex proprioceptive and exteroceptive somatosensory stimuli.

Seizures induced by exteroceptive somatosensory stimuli. Seizures induced by exteroceptive somatosensory stimulation are typically triggered by touching, tapping, rubbing, or pricking part of the body (98). A localized or regional hypersensitive trigger zone can often be defined. The seizures begin with sensory symptoms; a sensory Jacksonian seizure occurs, often followed by tonic motor manifestations, suggesting a supplementary motor area seizure. Subsequent generalization may occur. Consciousness is preserved, at least at the onset. Seizures induced by unexpected somatosensory (and other) stimuli are described in startle epilepsy.

Seizures induced by tapping, often by a single touch (“tap epilepsy”), manifest with brief reflex generalized myoclonic attacks associated with bilateral spike and wave EEG discharges (27; 155). They show marked similarities to the benign early infantile reflex absence seizures as illustrated with video-EEG recording by Voskuil (161). These typically occur in otherwise normal infants and toddlers. Also, tapping is a typical and common stimulus to elicit giant somatosensory evoked responses (without clinical response) in EEG in children with rolandic and other benign childhood focal epilepsies (150; 91; 119).

Seizures elicited by tapping or rubbing have been reported in some patients with cortical lesions localized in postrolandic and other regions (147; 87; 69), though normal brain MRI is more likely (80). Sala-Padro and colleagues have reported 4 adult patients with touch-induced seizures (130). They all had focal motor seizures after the stimuli, without impairment of awareness. Seizures were refractory to antiepileptic medication. The origin of such seizures was due to epileptic foci in the fronto-central regions associated with peri-central gyri lesions on MRI.

Toothbrushing-induced seizures are a rare form of reflex epilepsy; patients may have normal or abnormal brain MRI (67; 111; 90; 19; 63; 30). Small lesions of the primary somatosensory cortex may be seen.

Stimulation of the external ear conduct (auricular epilepsy) may also elicit seizures (132). Compulsive somatosensory self-stimulation inducing epileptic seizures has been documented in severe cases of infantile spasms (61; 149).

Diaper changing-induced reflex seizures, although very rare, are considered as a possible variant of sensory reflex seizures that represent a subtype of “rub-evoked reflex epilepsy”. A case of diaper changing-induced seizures has been described in baby with CDKL5-related epilepsy (142).

Hot water epilepsy. Hot water epilepsy, also known as “water immersion epilepsy” or “bathing epilepsy,” is detailed in the article on hot water epilepsy. Briefly, hot water epilepsy is characterized by epileptic seizures elicited by pouring hot water (40°C to 50°C) over the head (127; 09; 118; 100; 101; 157). Thus, this mode of reflex epilepsy requires a specific thermal cutaneous stimulus (ie, pouring of very hot water over the head). Less often, seizures may occur while using a shower, during tub bathing, or exceptionally, with cold water. Self-induction occurs in 10% to 20% of patients. Hot water-induced seizures are predominantly (80%) simple or complex focal with (25%) or without (75%) secondary GTCS. They may occur at any time during the bathing, even at the beginning of it. Their duration is usually 30 seconds to 3 minutes.

The ILAE Task Force on Classification and Terminology core group has introduced “hot water epilepsy in infants” as a new reflex epileptic syndrome with a relative confidence level of 2 (Table 1) (35), though age at onset widely varies from 2 months to 58 years with a mean of 13 years. Half of patients have their first seizure during the first decade of life, but in a third of patients, this happens after the age of 18 years. Male patients outnumber females in a ratio of 3:1.

Almost all cases of hot water epilepsy are seen in otherwise healthy children with normal neurologic examination and normal brain imaging. In a study from India of a large 4-generation family with autosomal dominant inheritance of hot water epilepsy, significant linkage was detected on chromosome 4q24-q28 (127).

Altering bathing techniques by lowering water temperature, showering, or sponging instead of pouring water over the head, and reducing the duration of the bath, are usually sufficient to achieve seizure control. Antiepileptic drug medication is only indicated when these stimulus-modifying or -preventing measures fail or when spontaneous seizures also occur. Intermittent oral administration of clobazam before a hot water bath has been found to be effective.

Contrary to the information above, some patients of a family segregating focal seizures and autistic spectrum disorder have reflex seizures triggered by bathing or showering. The underlying cause of this family’s disorder is a Q555X mutation of synapsin 1 (SYN1) on chromosome Xp11-q21 (110).

Appavu and associates described 2 boys with a unique form of bathing epilepsy characterized by the act of exiting out of water (05). The first patient had a family history significant for a brother with frontal lobe epilepsy. He underwent an evaluation in the epilepsy monitoring unit in which a reflex seizure was recorded while exiting the shower. This seizure was characterized by an ictal onset in the left frontal lobe and subsequent secondary generalization. The second patient initially had nonreflex seizures arising from the left temporal lobe and went on to develop reflex seizures on exiting water. For both patients, the precipitation of seizures was independent of water or environmental temperature, exposure of specific body parts, or duration of water immersion. Both children experienced a sensation of coldness, followed by convulsive or atonic activity.

Seizures induced by simple proprioceptive somatosensory stimuli. Simple proprioceptive stimuli (movement, including eye-closure, micturition, and defecation) have been well-described as precipitants of reflex epileptic seizures. However, genuine movement-induced (kinesigenic) seizures elicited by active or passive movements are rare (28; 121; 44; 94; 156; 13; 68; 146; 39). These should be meticulously differentiated from non-epileptic paroxysmal kinesigenic dyskinesia (83; 141).

Seizure-inducing movements can involve 1 particular joint or a group of joints, or eyes; gait (gait epilepsy) (74); and, exceptionally, simple ideation of the movement. Primarily, seizures are brief and focal sensorimotor. They mainly happen in patients with focal brain lesions in the parietal and sensorimotor cortex or supplementary motor area. A patient with focal cortical dysplasia (Taylor type) involving left motor cortex had both spontaneous and movement-induced (by arm wrestling) right upper limb right sensorimotor seizures with Jacksonian progression (146). Electroclinical seizure recording showed progressive increase of spiking over the left central, followed by recruiting rhythm and right tonic seizure. The patient became almost seizure-free after corticectomy, without neurologic sequelae (146). A case with video-EEG presentation was of a 78-year-old man who presented with abnormal right leg movements 2 days after mild traumatic brain injury (39). He displayed clonic movements of the right leg after a few seconds of voluntary movement of the same leg, lasting about 20 seconds. Brain MRI showed a small subarachnoid hemorrhage in the cerebral convexities. EEG at rest was normal, but demonstrated robust epileptiform discharges in the left frontotemporal region figure after voluntary movements of the right leg. Seizures subsided with clobazam 20 mg/d. Exceptional are seizures induced by micturition and/or defecation (143; 172; 52; 126; 01; 65; 162; 138; 129; 158). Seizures induced by defecation originate from the left temporal lobe (129).

Seizures induced by eating (eating epilepsy). Seizures induced by eating are characterized by seizures closely related to 1 or several aspects of eating (128; 86). Seizures usually occur in the early phase of eating (generally within the first 5 minutes) or within the first 10 minutes after eating. Rarely, patients have seizures at the sight, smell, or thought of food (33; 145); others may have them immediately after a heavy meal, suggesting gastric distension as a trigger (46). Seizures with eating are typically focal motor seizures, with or without auras or automatisms of temporolimbic type. They are almost always related to symptomatic epilepsy, although familial occurrence has also been reported (171). Two patients with left temporal lobe epilepsy were confirmed to have ictal asystole and eating epilepsy by video-EEG monitoring (06). Rarely, eating-induced epileptic spasms may occur (107). A boy carrying a maternally inherited MECP2 duplication had video-EEG confirmation of a cluster of eating-induced spasms, the only epileptic manifestation of the patient (26). Seizures induced by eating are usually associated with localized or regional EEG epileptiform activity, either from temporolimbic structures or from suprasylvian regions in association with larger lesions. EEG epileptiform activity that is generalized from the start is rare.

Rémillard and associates suggested that patients with temporolimbic seizures activated by eating have fewer spontaneous attacks and are more likely to have such attacks from the onset of their epilepsy than are patients with extralimbic, usually suprasylvian, seizure onset who have less constant activation by eating. Patients with suprasylvian seizure onset usually have more obvious extratemporal structural lesions and possible activation by specific thalamocortical afferents (128). They may also have seizures with other forms of buccal stimulation, such as toothbrushing or kissing. A prevalence of approximately 1 per 1000 to 2000 epileptic patients has been reported (159; 106; 86). Unusually high figures of eating epilepsy have been reported in 2 different areas of Sri Lanka by different investigators (136; 137). This could be attributed to inclusion criteria and ascertainment methods or related to genetic or ethnic factors and to the bulky meals rich in carbohydrates consumed by the patients (137).

von Stulpnagel and colleagues reported 8 patients with a SYNGAP1 mutation and chewing-/eating-induced reflex seizures as a new phenotype and compared them to other patients with eating epilepsy and genetic mutations (160). They presented the clinical and anamnestic features and retrospective analysis of video-EEG data of a 4-year-old index patient with SYNGAP1 mutation and chewing-/eating-induced seizures. Clinical and anamnestic features and home videos of 7 additional patients (4 female; ages 4-14 years) with SYNGAP1 mutation and eating-induced reflex seizures were compared. All reflex seizures of the index patient showed similar focal EEG pattern with 1 to 5 seconds high amplitude and irregular 3 Hz spike-wave complexes with initiation from left temporo-occipital, right temporo-occipital, or bi-occipital/temporo-occipital regions. Eyelid myoclonia was the most common seizure type in all 8 patients and it was typically initiated by eating or other simple orofacial stimuli. In the index patient eye closure preceded eating-induced eyelid myoclonia in the majority of the seizures (160). Ictal vomiting is a rare occasion of seizures induced by eating (148).

Seizures induced by auditory stimuli and musicogenic epilepsy. Seizures induced by sounds or words (audiogenic epilepsy) are rare (29). Musicogenic epilepsy is the commonest and best studied form (176; 48; 81; 163; 140; 122; 95; 102; 31; 37).

Musicogenic epilepsy is characterized by seizures induced by hearing certain sounds, typically music (25). Seizures have also been reported while the subject is exposed to the musical trigger during sleep or while merely thinking about it. The effective stimulus can be stereotyped for each patient and at times is exquisitely specific, but with no clear common pattern between patients. An affective component of the stimulus is evident in some patients, yet nonmusical sounds, such as whirring machinery, can be effective triggers in others. The seizures are of simple or complex focal type with interictal and ictal epileptiform activity recorded from either temporal region (135), usually the right. Most patients also have spontaneous seizures and reflex seizures that often begin over a year after the onset of spontaneous attacks (164). Musicogenic seizures have also been reported in infancy (93).

In a retrospective publication, there were 3 patients with musicogenic epilepsy out of a group of 1510 epilepsy patients (0.0019%) (37). Significantly, 2 of these patients had autoimmune epilepsy with antibodies against glutamic acid decarboxylase. Both patients had normal MRI, but FDG-PET showed medial temporal lobe hypometabolism (unilateral or bilateral) in both and also in the insula in 1 of them. Video-EEG documented onset of seizures from the right temporal lobe. The incidence of musicogenic epilepsy in patients with epilepsy and antibodies against glutamic acid decarboxylase was 2 of 22 (9%). The authors concluded that determination of antibodies against glutamic acid decarboxylase should be carried out in all cases of musicogenic epilepsy, even those with normal structural MRI.

Seizures induced by thinking. Seizures induced by thinking (“noogenic epilepsy”) occur in response to nonverbal higher cortical function and have been reported with a variety of stimuli, including arithmetic, drawing, playing cards or chess, decision-making, and solving Rubik's cube (166; 56; 04). These seizures do not typically appear to be activated by reading, writing, or by explicitly verbal tasks, but about 80% of patients are found to have more than 1 effective trigger. Seizures can be triggered in at least some of these patients without any real or contemplated movement of the hands, eg, by a task requiring a spoken answer to an orally presented arithmetic or spatial problem. Unlike primary reading epilepsy, patients with noogenic epilepsy have spontaneous seizures. The reflex and spontaneous attacks include bilateral myoclonus, absences, and generalized tonic-clonic seizures. Often these begin after a period of myoclonic jerks, but myoclonic jerks occurred without a following convulsion in 76% of patients reviewed by Andermann and colleagues, and 60% of patients had absence seizures often associated with myoclonic jerks (04). Not all patients had myoclonus, although some probably had juvenile myoclonic epilepsy. Myoclonic jerks and absence attacks may be ignored or unreported until a generalized tonic-clonic seizure occurs, and the patient then comes to medical attention. Seizures induced by thinking usually occur in the context of generalized epilepsy. Partial seizures and clear focal EEG abnormalities have been reported, but are the exception (109). The essential component in the seizure trigger appears to be nonverbal thought, the processing of numeric or spatial information, and possibly sequential decision-making. A third of the patients also show photoparoxysmal discharges, but not clinical photosensitivity. Studies provide more detail on the cerebral representation of calculation and spatial thought and document a bilateral functional network activated by such tasks (144). The parieto-occipital lobes might be partially involved in the neuronal network responsible for card game–induced reflex epilepsy (120).

Self-induced noogenic seizures have also been reported (88).

Seizures induced by acute emotional states are relatively uncommon (125). These are usually limbic seizures with psychic aura originating in networks involved in emotional process, with temporal lobe epilepsy being the most common, although without clear laterality or gender predominance (125).

Praxis-induced seizures. The term “praxis-induction” was introduced by Japanese authors who described seizures triggered by thinking about “complicated spatial tasks in a sequential fashion, [making] decisions, and practically responding by using a part of [the] body” (71). Writing is reported to be a major precipitating factor although reading is not (72). Hand or finger movements without “action-programming activity” (defined as “higher mental activity requiring hand movement” and apparently synonymous with praxis) are not effective triggers (Matsuoka et al 2000; Matsuoka et al 2005). Reflex upper-limb myoclonus occurs and may spread. This pattern occurs almost exclusively in juvenile myoclonic epilepsy (JME). It does not seem prominent in patients with thinking-induced seizures who do not also have prominent myoclonic reflex attacks. In its milder or most restricted forms, such as the morning myoclonic jerk of the arm manipulating a utensil, this phenomenon resembles cortical reflex myoclonus as part of a “continuum of epileptic activity centered on the sensorimotor cortex” (156). The motor component, either imagined or performed, is crucial in praxis-induction, whereas seizures induced by thinking are activated by tasks such as purely mental calculation of orally presented arithmetic tasks with no motor component in either the stimulus or the response. Like praxis-induced seizures involving the limbs, perioral reflex myoclonias with activation of the EEG, originally believed to be especially associated with reading epilepsy, are now recognized as typical of juvenile myoclonic epilepsy (99).

A patient presenting with praxis-induced myoclonic epilepsy at a late age has been reported, with video-EEG documentation (53).

Whether praxis induction and thinking induction involve 1 or 2 mechanisms has been thoroughly debated (73).

Startle-induced seizures and startle epilepsy. Briefly, startle epilepsy is characterized by seizures triggered by unexpected sudden sensory stimuli, usually sound or touch. Most patients have intractable seizures, evident static encephalopathy, and neurologic deficits (infantile hemiplegia is common) with corresponding abnormal imaging. The startle response consists of axial tonic posturing, frequently causing falls, which can often be traumatic. Concurrent symptoms, such as marked autonomic manifestations, automatisms, laughter, and jerks, may occur. Less commonly, startle-induced seizures may be atonic or myoclonic, particularly in patients with cerebral anoxia. Seizures are frequent, occurring many times a day, and sometimes progress to status epilepticus. The prognosis is often poor, particularly for those with severe preexisting encephalopathies. Differentiation between startle-induced seizures and exaggerated startle responses can be challenging. Startle epilepsy should be differentiated from other nonepileptic disorders with abnormal responses to startling events and, mainly, hyperekplexia. Differentiation between seizures and exaggerated startle responses can be challenging. Startle epilepsy is often resistant to medical treatment. Epilepsy surgery may be beneficial to those with focal etiology of seizures.

Self-induced seizures. Self-induction has been well established as a mode of precipitation, but prevalence is disputed from a small number to as high as 30% of photosensitive patients. Self-induction is employed not only by the mentally handicapped, as it was initially reported, but also by patients of normal or above average intelligence. Self-induction is more common in patients with photosensitive epilepsy, but it is also observed in other types of reflex epilepsy such as hot water epilepsy. Patients get indescribably immense pleasure out of this seizure, which is usually an absence seizure and less often a myoclonic jerk. GTCS may occur as the result of this self-induced intermittent photic stimulation, but this is probably an unwanted effect. Sexual pleasure and masturbation are exceptional. “It gives me the same pleasure as when I get a most precious Christmas gift” and “I love it, a release of tension” are common confessions by the patients. Spontaneous seizures are not pleasurable, which indicates that the ictal enjoyment is due to the seizure-seeking behavior and maneuvers employed for self-stimulation.

Prognosis and complications

Prognosis for these different seizure types depends on the underlying disorder and on the degree to which stimuli can be modified or avoided, or the effect of treatment. Some patients may have only 1 or a few reflex seizures that can be entirely avoidable (eg, video game–induced seizures). In others, reflex seizures are controlled by avoiding precipitating factors and/or appropriate prophylactic medication (eg, reading epilepsy). Other reflex seizures may be lifelong and intractable (eg, progressive myoclonic epilepsies and most startle-induced seizures).

Complications are similar to those for other similar spontaneous seizures.

Biological basis

Localization

Localizations and regional or functional triggers of reflex seizures are summarized in Tables 3 and 4. The topography of physiological networks involved follows the posterior-to-anterior trajectory of brain development, reflecting age-related changes in brain excitability (85).

Photosensitive and pattern-sensitive reflex epilepsy involves the primary visual occipital cortex (114; 12). Proprioceptive-induced seizures classically involve the rolandic sensorimotor area of the hemisphere contralateral to the clinical seizure onset (36). Reading-induced focal seizures with alexia seem to originate from the dominant supramarginal and angular gyrus (89; 47). Conversely, in primary reading epilepsy localization may be in deep cortical and subcortical structures, in particular the left frontal piriform cortex (154). Musicogenic seizures commonly originate from the right temporal lobe (164; 49; 48).

Etiology and pathogenesis

The vast majority of reflex epileptic seizures are genetically determined, but others are due to fixed or transient neurologic deficits (76; 77).

Italiano and associates published a comprehensive review on the role of genetic determinants in humans and in animal models of reflex seizures and epilepsies (77). The main finding was that autosomal dominant inheritance with reduced penetrance was proven in several families with photosensitivity. Molecular genetic studies on EEG photoparoxysmal response identified putative loci on chromosomes 6, 7, 13, and 16 that seem to correlate with peculiar seizure phenotype. Autosomal dominant inheritance with incomplete penetrance overlapping with a genetic background for idiopathic generalized epilepsy was proposed for some families with primary reading epilepsy. Musicogenic seizures usually occur in patients with focal symptomatic or cryptogenic epilepsies, but they have been reported in rare genetic epilepsies such as Dravet syndrome. A single LGI1 mutation has been described in a girl with seizures evoked by auditory stimuli. Eating seizures usually occur in symptomatic epilepsies, but evidence for a genetic susceptibility was mainly provided by a family report from Sri Lanka. Eating seizures were also reported in rare patients with MECP2 duplication or mutation. Hot water seizures are genetically heterogeneous, but 2 loci at chromosomes 4 and 10 were identified in families with likely autosomal dominant inheritance. Startle-induced seizures usually occur in patients with symptomatic epilepsies, but have also been reported in the setting of chromosomal disorders or genetically inherited lysosomal storage diseases. The authors concluded that the genetic background of reflex seizures and epilepsies is heterogeneous and mostly unknown, with no major gene identified in humans.

Pathophysiology

Two types of study in animal models of reflex seizures are: (1) the study of irritative cortical lesions and their activation by specific stimuli and (2) the study of naturally occurring reflex epilepsies or seizures induced by specific sensory stimulation in genetically predisposed animals. The first approach has been used by Italian physiologists since 1920 (see Historical note and nomenclature). The second experimental approach, the study of naturally occurring or induced reflex seizures in genetically susceptible animals, has been pursued in various animal models, including primates (Menini and Silva-Barrat 1998; 108).

Little is known concerning the pathophysiology of human photosensitivity to intermittent photic stimulation. However, there is a considerable body of evidence concerning pattern sensitivity, and it may be presumed that similar mechanisms are involved (12; 175; 38). They can also be applied to other reflex epilepsies and seizures and may represent a common mechanism of reflex seizure genesis. Recruitment of a “critical mass” of epileptogenic cortex in response to the reflex seizure stimulus can result in epileptiform EEG activity or a clinical seizure. This recruitment can be understood as relatively direct in some, but in other cases may involve the participation and interaction of several cortical areas or of cortex and subcortical structures activated by the external trigger. Other factors may have the opposite effect, reducing the amount of involved cortex and the likelihood that a seizure will occur.

The model of human pattern-sensitive epilepsy, reviewed by Radhakrishnan and colleagues (123), is of special interest because it shows that generalized clinical events and generalized or bilateral EEG abnormalities can be activated by a specific functional stimulation with a known localization in the case of pattern-sensitive epilepsy involving hyperexcitability of primary visual cortex (12; 134). This electroclinical pattern is found in many subjects with several types of reflex seizure triggers and idiopathic generalized epilepsy without neurologic deficit or evident lesions on imaging, and who are, thus, presumed to have diffuse cortical hyperexcitability with a genetic component.

Seizures induced by thinking and by “praxis” epilepsy appear to follow this model. Like these, the dissociation between drawing and writing-induced seizures discussed by Kho and colleagues suggests the existence of verbal and nonverbal hyperexcitable networks to induce these attacks (84). Photosensitive occipital partial seizures also occur in patients with idiopathic generalized epilepsy (60), and motor activity can elicit seizures in nearly 50% of patients with juvenile myoclonic epilepsy (97), an “idiopathic generalized epilepsy.” The pattern of praxis induction fits well with juvenile myoclonic epilepsy, in which there is apparent regional hyperexcitability of sensorimotor cortex within an apparently generalized epileptic disorder (169) and possibly relevant morphologic abnormalities involving the motor system as well (168). These and other observations in both reflex and spontaneous epileptogenesis suggest that the postulated diffuse cortical hyperexcitability in idiopathic generalized epilepsy is not necessarily uniform; specific activities can activate specific cortical systems or functional networks spread over several cortical regions in 1 or both hemispheres and produce focal or regional discharges, or partial seizures, which may generalize. This does not invalidate a diagnosis of underlying generalized epilepsy, but shows that the biological substrate of generalized epilepsy can be complex (Binnie 2005; 38). As predicted by Clementi’s model, focal or regional brain lesions can also be associated with reflex seizure triggering by appropriate stimuli. Such patients may also have spontaneous seizures and many have neurologic deficit and abnormal imaging. Prior to MRI, many malformations of cortical development were undetected, but these can be responsible for reflex seizures, especially drop attacks or axial myoclonus. Selected cases may be treated surgically (113).

The pathophysiology of reading epilepsy may be different in those with the jaw myoclonic type and in those with the focal seizures of alexia (89; 07; 47; Op de et al 2011; 154). The focal type with alexia is a pure reflex focal epilepsy with the stimulus (reading) directly acting on and disturbing the localized brain region subserving reading in the dominant occipitotemporal junction, which coordinates information that is gathered from visual and auditory processing and assigns meaning to the reading. These include the dominant supramarginal gyrus, which seems to be involved in phonological and articulatory processing of words, and the angular gyrus, which, together with the posterior cingulate gyrus, seems more involved in semantic processing.

Proprioceptive-induced seizures classically involve the rolandic sensorimotor area of the hemisphere contralateral to the clinical seizure onset. Maximum EEG electronegativity appears to be at the central vertex electrode in a published EEG record of seizures induced by walking (74). The supplementary motor area may also be involved. This localization has been confirmed by imaging and intensive monitoring. Cerebral lesions are often evident and may have occurred well before the onset of attacks. Human proprioceptive-induced seizures are seen during nonketotic hyperglycemia, in which attacks are induced by active or passive limb movement and, more rarely, with conjugate gaze (15; 32); an associated neurologic deficit related to a remote lesion may be transiently unmasked during the period of seizures (15). Acute cerebral lesions or diffuse encephalopathies may also be accompanied by self-limited proprioceptive-induced seizures.

Proprioceptive-induced seizures have been studied with a chronic alumina focus in 1 foot area of the monkey brain. Reflex seizures were elicited by stimuli that produced proprioceptive input to the hyperexcitable cortical area. Seizures could not be elicited after curarization (18). Transcortical reflex loops appear to be involved in generating the proprioceptive-induced motor response in the cat with a penicillin focus (51).

Many patients with eating epilepsy have seizures that can be activated only by obvious combinations of stimuli (40), and alerting stimuli have been reported to abolish attacks (45), providing at least circumstantial evidence for involvement of an increasing cortical mass and of subcortical influences that may promote or inhibit seizure occurrence in some cases of reflex epilepsy. It appears that localization of seizure onset and the nature of the seizure trigger are related in these attacks. Seizures in patients with suprasylvian lesions may be triggered by other oral activities and may represent a particularly noticeable type of seizure induced by proprioceptive or by somatosensory stimulation. They may be different from patients with temporolimbic-onset seizures, in whom taste and autonomic afferents may play a more important role and in some of whom seizures may also be related to emotional or autonomic components of eating or to gastric distension, with possible participation of limbic and autonomic afferents. In a case of eating epilepsy studied using simultaneous EEG-fMRI, seizures were triggered by the sight or smell of food, and frontotemporal discharges were noted on video-EEG (131). Similarly, frontotemporoparietal involvement was noted on SPECT and EEG-fMRI.

The pathophysiology of musicogenic epilepsy was initially attributed to conditioned response (139), but this view was not accepted (43). Chronic temporal lobe depth electrode studies in epileptic subjects without musicogenic epilepsy suggest different lateralizations for different components of a musical stimulus (165). Creutzfeldt and Ojemann confirmed that musical stimuli may have widespread effects on neuronal activity in human temporal lobes extending well beyond the rather restricted primary auditory area (24; 92), that different components of music have different effects, possibly with specialized lateralization and localization, and that the effects of music are different from those of speech. PET studies in patients and others show predominant involvement of right hemisphere structures in networks involved in processing musical information, extending well beyond the classical auditory cortex of Heschl’s gyrus (79). Studies in subjects with musical hallucinations show that the primary auditory cortex is not “a sufficient substrate for higher-order pattern perception” (58).

The primate auditory cortex is considered to consist of a central core of primary cortex that receives thalamic projections and is linked to several “belt” areas. Primary cortex has multiple tonotopically organized sections and is especially sensitive to pure tones. Belt regions show more sensitivity to complex stimuli and are less tonotopically organized (79). Zifkin and Zatorre also note that more complex musical processing tasks activate more cortical and subcortical territory bilaterally but with right hemisphere predominance (176). Zatorre and coworkers provide a review of auditory-motor interactions in music perception and production (173). Thus, hyperexcitable cortical areas could be stimulated to different degrees and extents by different musical stimuli in patients sensitive to musical triggers. Gloor suggested that responses to limbic stimulation in epileptic subjects depend on widespread neuronal matrices linked through connections that have become strengthened through repeated use, of interest in considering the delay from seizure onset to the development of sensitivity to music (54). The heterogeneity of musicogenic epilepsy and the several different temporal lobe foci reported to be responsible for these seizures can perhaps be understood as representing sensitivities to different parts of a musical or auditory stimulus (152). The finding of LGI1 mutations in lateral temporal epilepsy triggered by auditory stimulation and telephone-induced reflex seizures (see above) is just 1 type among many genetic-clinical-electrographic patterns that may exist.

Italiano and colleagues reviewed available knowledge about reflex seizures with special emphasis on the difference between "generalized" reflex seizures induced by visual stimuli, thinking, praxis, and language tasks and "focal" seizures induced by startle, eating, music, hot water, somatosensory stimuli, and orgasm (76). Evidence from animal, clinical, neurophysiological, and neuroimaging studies is supporting the concept that "generalized" reflex seizures, usually occurring in the setting of idiopathic generalized epilepsy, should be considered as focal seizures with quick secondary generalization through corticoreticular or cortico-cortical pathways.

Epidemiology

Reflex seizures affect patients of all ages and have a prevalence of 4% to 7% among patients with epilepsies.

Prevention

Avoidance of precipitating factors, such as those discussed in Management, is a significant part of prevention.

Differential diagnosis

Seizure phenomenology

The differential diagnosis firstly involves the distinction of genuine epileptic seizures from nonepileptic paroxysmal attacks (118).

Eyelid myoclonia induced by eye closure and jaw jerks of primary reading epilepsy are often misdiagnosed as ticks. Reflex and spontaneous visual seizures are commonly misdiagnosed as migraine with visual aura. Conversely, paroxysmal kinesigenic dyskinesia is often misdiagnosed as movement-induced epileptic seizures. Hyperekplexia is often misdiagnosed as tap-induced seizures.

The association of the epileptic seizures with a particular triggering/precipitating stimulus and other facilitating factors should be thoroughly established. For example, eye closures inducing eyelid myoclonia should not be equated with attempts for self-induction, which is a common error.

Underlying disorders

Seizures triggered by visual stimuli are seen in several epilepsy syndromes (Tables 1 and 2). These include recognized syndromes of reflex epilepsy such as idiopathic occipital photosensitive epilepsy, primary reading epilepsy, and hot water epilepsy; unrecognized syndromes of reflex epilepsy such as Jeavons syndrome; syndromes commonly manifesting with reflex epileptic seizures such as juvenile myoclonic epilepsy, Dravet syndrome, and progressive myoclonic epilepsies; and reflex seizures from focal structural lesions.

Diagnostic workup

Diagnostic workup is required for patients with spontaneous seizures of the same type. Properly gathered clinical information is fundamental in establishing with reasonable confidence whether the seizures are truly of epileptic origin, their type(s) (focal or generalized), and the underlying syndrome or cause. Subsequently, EEG and brain imaging are the most valuable tests to confirm diagnosis and cause.

In reflex epileptic seizures, the discovery of the triggering stimulus is often, but not always, straightforward. Photosensitivity is commonly revealed in routine EEG with appropriate intermittent photic stimulation.

However, EEG photoparoxysmal responses should not be equated with clinical photosensitivity, though this is often the case. As with all other aspects of EEG, intermittent photic stimulation–induced abnormalities should be meticulously correlated with the clinical manifestations; the one complements the other. Furthermore, routine EEG does not usually examine other modes of reflex-seizure precipitation. These, such as reading, somatosensory, and others, should be properly applied and tested to delineate the trigger when the clinical evidence points to them (150; 91; 119).

These are of eminent clinical significance because of probably effective or at least helpful nondrug management, such as stimulus avoidance or modification.

Brain MRI is mandated for those with suspected brain lesions of structural reflex epilepsy. High resolution MRI should be used because subtle cortical dysplastic or other lesions may be missed or may be found only in a surgical pathology specimen (96).

Management

Avoidance, prevention, or modification of the provocative stimulus. Avoidance, prevention, or modification of the provocative stimulus may be sufficient for many patients with reflex seizures. The usual advice about avoiding sleep deprivation, excessive alcohol use, and recreational drugs should be given.

In pure photosensitive and pattern-induced seizures, avoidance of the provocative stimulus may be adequate, eg, patients with television-induced seizures should be advised to view the television in a well-lit room, maintain a maximum comfortable viewing distance (typically > 2.5 m for a 19-inch screen), use the remote control and, if it is necessary to approach the screen, cover 1 eye with the palm and avoid prolonged watching, particularly if sleep deprived or tired. Occlusion of 1 eye is also advised when photosensitive individuals are suddenly exposed to flickering lights such as in nightclubs. Patients with video game–induced seizures should attempt to go without video games, or the time of playing should be significantly restricted and confined to periods when patients are not sleep deprived. Conditioning treatment or wearing appropriately tinted glasses has been recommended. A commercially available blue lens, Z1, was found to be highly effective in controlling photoparoxysmal discharges. An optometric technique, colorimetry, enables the perceptual effects of ophthalmic tints to be evaluated subjectively, optimized, and then prescribed in tinted spectacles (Wilkins 1995a; 167).

Modification of reading and talking habits may be successful in reading epilepsy.

Altering bathing techniques, eg, lowering water temperature, showering, or sponging instead of pouring water over the head, and reducing the duration of the bath are usually sufficient to achieve seizure control in hot water epilepsy.

Prophylactic antiepileptic medication. Prophylactic antiepileptic drugs are needed when triggering stimuli are not preventable, when seizures are frequent, and/or when seizures occur spontaneously. The choice of an AED depends on its specific efficacy on the type(s) of reflex and spontaneous seizure and the particular epileptic syndrome (118; 76). This is no different from treatment of spontaneous seizures of the same type.

In photosensitive, pattern-induced seizures and those with syndromes of generalized epilepsies, valproate, levetiracetam, lamotrigine, and clonazepam suppress photosensitivity, in that order of efficacy. Valproate is effective for all types of photically or spontaneously induced seizure. Levetiracetam is also effective for all types of photically or spontaneously induced seizures; it is much more efficacious in myoclonic seizures and GTCS than absence seizures. Lamotrigine is effective for all but myoclonic types of photically or spontaneously induced seizure. Its efficacy in photoparoxysmal discharges has been studied, mainly with valproate1, and may be due to their pharmacodynamic interaction. Lamotrigine may be particularly useful in absence seizures when valproate is ineffective or undesirable. Clonazepam is effective in photically or spontaneously induced myoclonic seizures, but is ineffective against and may exaggerate GTCS. Most of the other AEDs (ie, carbamazepine, gabapentin, oxcarbazepine, phenytoin, pregabalin, tiagabine, and vigabatrin) are contraindicated because they are ineffective (ie, they induce side effects without providing any therapeutic benefit) in addition to depriving patients of appropriate treatment. Self-induced visual-evoked seizures may be difficult to treat. The attacks may be pleasurable, and there is clear secondary gain for some patients (11; 151).

In primary reading epilepsy, clonazepam 0.5 to 1.0 mg at night is highly effective.

In startle epilepsy, there is no established drug of choice, and therapy is often unsatisfactory. Clonazepam, clobazam, and carbamazepine are frequently used. The role of the newer AEDs in startle seizures has not been investigated. Lamotrigine and levetiracetam have been found to be very effective in small series of patients.

In hot water epilepsy, intermittent oral administration of clobazam before a hot water bath has been found to be effective. Withdrawal of medication should start on remission of seizures.

Neurosurgery. When focal seizures occur, medically intractable cases should be recognized early and assessed for surgical treatment.

Outcomes

Outcomes vary significantly depending on the type and cause of reflex seizures. Some patients may have only 1 or a few reflex seizures in their life, whereas others may suffer frequent seizures that are intractable to treatment.

Special considerations

Pregnancy

Women on antiepileptic medication, particularly valproic acid, need proper advice prior to conception.