Differential diagnosis between epileptic seizures and psychogenic nonepileptic seizures based on semiology

Seizures can be divided into three major categories: epileptic seizures (ES), psychogenic nonepileptic seizures (PNES), or physiologic nonepileptic events. The majority of patients with recurrent seizures are initially presumed to have epilepsy and are treated with antiepileptic drugs. PNES present as paroxysmal time-limited, alterations in motor, sensory, autonomic, and/or cognitive signs and symptoms, but are not caused by ictal epileptiform activity [1]. PNES are diagnosed in at least 10–40% of the patients seen for long-term monitoring of epilepsy, and are as disabling as epilepsy [2]. Patients with long standing PNES, approximately 10–15% turn out to have additional epilepsy [3]. The manifestations of PNES often resemble those of ES, and accurate distinction among these conditions is crucial because treatment choice depends on it.

The gold standard test would be the synchronous recording of all events of consideration by video, heart rhythm by electrocardiogram, and electroencephalogram (EEG), and exclude other causes [4]. Thus, it implies accessibility to a monitoring unit with specialized reviewers and services. Nonetheless, video-EEG (vEEG) monitoring not infrequently fails to capture the events, and it may not differentiate certain types of frontal lobe epileptic seizures from PNES. Besides, vEEG is not without error, and may not be entirely independent of semiology, as its interpretation may partially rely on information gathered from seizure videos. Other limitations include that vEEG is expensive and is not readily available at many centers. Pattern recognition of events forms the cornerstone of interpreting vEEG findings. For these reasons, we decide to focus on semiological characteristics differentiating PNES from ES.

PNES pose clinical challenges in terms of diagnosis. Recent studies have thrown further light on the extent to which features of PNES semiology may distinguish PNES patients from those with epilepsy. It’s worth noting that single sign is unreliable as diagnostic discriminator, while in clinical practice even informed specialists contextualize multiple signs to hypothesize seizure etiology. Clusters of semiological elements may differentiate PNES more clearly from ES [5].

PNES mimic different types of ES, and a classification of PNES would be useful to increase diagnostic accuracy. However, an internationally accepted classification of PNES is not yet available. Several studies have aimed to identify homogeneous groups of PNES based on specific combinations of clinical symptoms and signs, but there is no uniformity of nomenclature.

Magaudda et al proposed a clinical classification of PNES, which classified seizures into four classes: (1) Hypermotor, (2) Akinetic, (3) Focal Motor, and (4) Subjective Symptoms [6]. Seizures mainly characterized by either tonic, clonic, or dystonic generalized movements were considered as belonging to class Hypermotor in correspondence with convulsive ES. Seizures mainly characterized by unresponsiveness and absence of movements were assigned to class Akinetic, corresponding to atonic or absence ES. Seizures with focal motor movements were assigned to Focal Motor in correspondence with epileptic partial motor seizures. Subjective Symptoms seizures were those mainly characterized by experiential phenomena reported by the patients. Though the inter-rater reliability for current clinical classification of PNES is only moderate [7], we adopt the criteria suggested by Magaudda et al to subdivide common symptoms and signs. They proposed four classes corresponding to the most frequently described ones in clinical practice, and classified PNES by considering the most prominent seizure characteristics. It’s the first study in which the validity of a new PNES classification was established and reached in subjective level and an automatic analysis framework, as provided by a machine learning (ML) approach.

Semiology provides important diagnostic information, though it’s unlikely that a single sign will prove sensitive and specific enough to distinguish PNES from ES. Prospective evaluation of combinations of these signs could lead to the development of validated clinical models.

One limitation is the lack of a standard classification of semiological groups among PNES. As suggested by Magaudda et al, a classification considering too many clinical signs (especially if they were not specific to one class) could cause confusion. To date, most studies concentrated mainly on motor signs, while information of PNES presenting with subjective symptoms or unresponsiveness is scarce. Classes primarily based on motor signs, however, presented the worst rate of interrater agreement. For example, Seneviratne et al classified PNES into six categories: (1) rhythmic motor PNES; (2) hypermotor PNES; (3) complex motor PNES; (4) dialeptic PNES; (5) nonepileptic auras characterized by subjective sensations; (6) mixed PNES [32]. The disagreement concerned more frequently “rhythmic”, “complex”, and “hypermotor” PNES. Contrary to ES, development of PNES does not correlate with an anatomic organization, often fluctuating, and is difficult to characterize. Furthermore, in cases of prolonged PNES, motor signs can evolve from one class to another.

Establishing the PNES diagnosis requires a clear understanding of the target events through not only the patient’s report, but also what other family members, friends or caregivers have observed. One challenge is that seizure witness provide unreliable accounts for seizure semiology. It is proven that discriminating one type of seizure from another is a learnt skill and requires neurologic training [33]. Acquisition of home video recordings of seizures may lessen the need to rely on eyewitness reports, and should be encouraged.

The situation becomes more complex due to the fact that epilepsy is a recognized risk factor for the development of PNES. PNES occur in 5–20% of adult populations with epilepsy and 15–40% of adult referral to epilepsy centers. Differential diagnosis can be challenging without vEEG. That’s partly because ES antedate, occur simultaneously with PNES, or even PNES appear as the initial presentation [34]. In practice, the diagnosis is often iterative, rather than a “one shot” process, and long-term clinical follow up may be important.

Some symptomatic differences help to distinguish ES from PNES, but need to be analyzed by combination of these symptoms and signs, long-term follow-up is sometimes required.