Epilepsy in Parry–Romberg syndrome and linear scleroderma en coup de sabre: Case series and systematic review including 140 patients

Abstract

Parry–Romberg syndrome (PRS) and linear sclerosis en coup de sabre (LScs) are rare, related, autoimmune conditions of focal atrophy and sclerosis of head and face which are associated with the development of focal epilepsy. The scarcity of PRS and LScs cases has made an evidence-based approach to optimal treatment of seizures difficult. Here we present a large systematic review of the literature evaluating 137 cases of PRS or LScs, as well as three new cases with epilepsy that span the spectrum of severity, treatments, and outcomes in these syndromes. Analysis showed that intracranial abnormalities and epileptic foci localized ipsilateral to the external (skin, eye, mouth) manifestations by imaging or EEG in 92% and 80% of cases, respectively. Epilepsy developed before external abnormalities in 19% of cases and after external disease onset in 66% of cases, with decreasing risk the further from the start of external symptoms. We found that over half of individuals affected may achieve seizure freedom with anti-seizure medications (ASMs) alone or in combination with immunomodulatory therapy (IMT), while a smaller number of individuals benefitted from epilepsy surgery. Although analysis of case reports has the risk of bias or omission, this is currently the best source of clinical information on epilepsy in PRS/LScs-spectrum disease. The paucity of higher quality information requires improved case identification and tracking. Toward this effort, all data have been deposited in a Synapse.org database for case collection with the potential for international collaboration.

1. Introduction

Parry–Romberg syndrome (PRS), also called progressive hemifacial atrophy (PHA), is a rare disease characterized by unilateral atrophy of skin, eye, subcutaneous tissue, muscle, and bone of one side of the face and scalp, and, uncommonly, trunk and extremities [1]. Parry–Romberg syndrome is a variant of craniofacial scleroderma often considered to exist on a spectrum with linear scleroderma “en coup de sabre” (LScs), a form of localized scleroderma of the paramedian face that causes an individual to appear as if scarred by the literal “strike of a sword” [1,2].

Parry–Romberg syndrome is named after early 19th century physicians Caleb Hillier Parry and Moritz Heinrich Romberg, who independently characterized the condition [1]. The term progressive hemifacial atrophy (PHA, or facial hemiatrophy) describing the same disease process was later coined by Dr. Albert Eulenburg in 1871 [1]. It is not a modern disease, however, as evidence for its existence has been found in archeological studies of Egyptian mummies as old as the first century Common Era [3]. While this disease entity has been a topic of recorded research for nearly 200 years, significant progress in understanding its pathophysiology has yet to be made because it occurs so infrequently [4]. Parry–Romberg syndrome and LScs conditions are typically self-limiting and may progress for years to decades before apparently extinguishing [1]. Some cases can involve the tissues of the eye, resulting in enophthalmos and visual symptoms, or the mouth, resulting in unilateral tongue atrophy and malocclusion [1,2]. Uncommonly, the ipsilateral neck, thorax, extremities, or entire hemibody can be involved, and, rarely, PRS or LScs can develop bilaterally [1,2].

Important to neurologists, persons with PRS and LScs frequently report significant neurological issues, including headaches/migraine, facial pain, and epilepsy [5,6]. The pervasiveness of neurological sequelae in PRS and LScs suggests a common underlying immunologic mechanism for neurological and external findings.

Parry–Romberg syndrome and LScs are known to cause focal epilepsy of either inflammatory/autoimmune or structural etiology [7,8]. Due to the sporadic nature of cases, understanding of risk factors, timing of onset, and outcomes of epilepsy in PRS and LScs is poor, and there have been no randomized clinical trials on management of this population. Case reports and series currently offer the best way to understand epilepsy in PRS and LScs. Our goal was to study the larger context of epilepsy in PRS and LScs after encountering three cases of PRS or LScs with epilepsy of varying severity at our institution. Here we present these cases and a systematic review of the literature of epilepsy in PRS and LScs to characterize seizure presentations, treatments, and outcomes.

2. Methods

2.1. Institutional chart review

Case series investigation was approved by the Institutional Review Board (IRB) at the University of Colorado Hospital (UCH). University of Colorado Hospital EPIC charting system was search using SlicerDicer from January 1st, 2010 to December 6, 2020 for cases of PRS (disorder of the facial nerve, ICD-10 G51.8) or LScs (ICD-10 L94.1) with a concomitant diagnosis of seizure (ICD-10 G40.89) or epilepsy (ICD-10 G40). Patient data were reviewed to confirm the diagnoses and then deidentified for analysis and publication. International CAse REporting (CARE) guidelines were followed for case presentations [9]. Details extracted included epilepsy history, relationship to PRS or LScs disease onset, diagnostic findings, treatments, and outcomes.

2.2. Systematic literature review

The international Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) protocol for systematic literature reviews were adhered to for the literature search [10]. The search was tailored toward finding all articles on PRS, LScs, PHA, and linear scleroderma (LS) also describing epilepsy manifestations, treatments, and outcomes. Searches were conducted across all dates in PubMed, Medline, ClinicalKey/Elsevier, Embase, BioMed Central, EBSCO, and Cochrane. The search phrase used for PubMed, Medline, BioMed Central, and Cochrane was: ((Parry–Romberg syndrome) OR (hemifacial atrophy) OR (linear scleroderma) OR (en coup de sabre)) AND ((seizures) OR (epilepsy)). This search phrase was too broad for ClinicalKey/Elsevier, Embase, EBSCO, so the search phrase was broken into four individual searches starting with each rheumatological diagnosis followed by seizures or epilepsy for these databases, for example: (Parry–Romberg syndrome) AND (seizures) AND (epilepsy).

Article titles and abstracts were screened for inclusion of reference to PRS, PHA, LS, or LScs with or without a mention of seizures or epilepsy. Articles were excluded if the main diagnoses given in the title/abstract were other conditions, e.g., Dyke–Davidoff–Masson syndrome, etc., or if skin changes were clearly linked to drug exposure, e.g., valproate. Non-peer reviewed publications such as conference proceedings and abstracts were excluded. Included articles were then retrieved as PDFs from the online database or requested from the University of Colorado Strauss Health Sciences Library. After full text review, articles were excluded if they did not report cases with a diagnosis of PRS, PHS, LScs, or LS spectrum disease with a simultaneous diagnosis of seizures and epilepsy or if individualized patient data could not be gleaned from the text. Cases describing seizures with localized linear or plaque scleroderma of the body (non-craniofacial distribution) were included for analysis. Non-English articles were translated using Google Translate. Studies performed as retrospective chart reviews over a specified time period for all individuals with PRS/LScs-spectrum disease, with or without epilepsy, were considered to be of “higher methodological quality.”

Data extracted were type and location of external findings, ages of onset of epilepsy and external findings, seizure characteristics, EEG and imaging findings, epilepsy treatment and outcomes. In addition, the journal specialty was recorded to evaluate for possible reporting bias related to the journal’s focus. Numerical and descriptive information from each study was entered into an Excel spreadsheet for quantification. Data were graphed as percent of the given total. Epilepsy onset was determined by subtracting the reported age of first seizure from the earliest report of skin disease, in years. Epilepsy diagnoses given in the publication were translated into the current diagnostic terminology as specified by the International League Against Epilepsy (ILAE) [7]. Some analyses were repeated on the higher methodological quality publications to assess for bias in the results with the entire dataset.

2.3. Data availability statement

Data collected from the literature search were uploaded to a publicly available data repository in Synapse.org at http://www.synapse.org/EpilepsyinPRSandLSCS. Deidentified patient data were uploaded to a private data repository also stored at Synapse.org. Specific instructions on data curation and use can be found on the project Wiki (see Supplemental Material for instructions on database navigation).

3. Results

3.1. Cases 1–3

Retrospective review of medical records in the UCH hospital system identified 23 cases of PRS and eight cases of LScs out of 5,583,242 patients. Calculated prevalence for these conditions combined was 0.6 cases per 100,000 people. Of these cases, 26 were in females and five were in males, giving a female-to-male ratio of 5 to 1. Five cases (16%) of PRS or LScs carried a diagnosis of seizures or epilepsy. Three cases had full records available and are reported below; two cases were seen at associated facilities with only limited records accessible and were excluded.

3.1.1. Case 1

This patient is a right-handed woman who presented at age 20 years to the neurology clinic due to new onset seizures. Her first seizure was experienced out of sleep and consisted of tonic-clonic (GTC) convulsions followed by postictal right-hand weakness. She was started on levetiracetam with no recurrence of convulsions. However, she continued to experience daily seizures with an aura of dizziness followed by eye-fluttering and loss of awareness. She was transitioned first to 100-mg topiramate, and then – due to adverse effects – to 400-mg lamotrigine and has remained seizure free on either drug for eight years after initial presentation. MR imaging after the first seizure showed T2/FLAIR hyperintensities in the left periatrial white matter, and four years later minimal lesion expansion into the temporal white matter and occipital cortex (Fig. 1A). Interictal EEG was normal.

Fig. 1.

Brain MRI images for patients with PRS and LScs seen at the University of Colorado. (A) Case 1, axial T2/FLAIR MR image showing left parieto-occipital gray and white matter hyperintensities. The big yellow arrow points to the putative seizure focus within the gray matter of the parieto-occipital lobe posteriorly. The thin white arrow indicates the region affected by left-sided linear scleroderma en coup de sabre (LScs) that received cosmetic treatment with injectable filler. (Bi) Case 2, axial T2/FLAIR MR image showing left-sided brain hemiatrophy with frontal hyperintensities in a patient with left-sided Parry–Romberg syndrome (PRS). The red arrow indicates atrophic hand knob of the pre-central gyrus likely relating to the patient’s right hand weakness. (Bii) Case 2, coronal T2/FLAIR MR image showing left hemiatrophy with left hippocampal sclerosis representing possible seizure focus (yellow arrow). (Ci) Case 3, axial T2/FLAIR MR image showing right hemiatrophy with medial temporal sclerosis in a patient with right-sided PRS. Note the facial asymmetry and enophthalmos characteristic of the patient’s PRS; (Cii) Case 3, axial T2/FLAIR MRI image showing hyperintensities in the foot area of the pre-central gyrus (yellow arrow). The white horizontal line indicates the imaging software scout line for the location of coronal image. (Ciii) Case 3, coronal T2/FLAIR MRI image showing extensive right-sided gray and white matter hyperintensities. The yellow arrow points to the foot area of the pre-central gyrus found to be one of several epileptogenic foci in the right hemisphere. The right side is indicated by the letter “R” in each image.

Dermatologic findings were first noticed one year and six months after onset of seizures when the patient developed a vertical band-like thickening of the skin over the left forehead that progressed to skin indentation. She was diagnosed with LScs and treated with 25-mg subcutaneous methotrexate for two years with no further progression of skin findings. Early after her initial skin diagnosis, she was also treated with two weeks of pulse steroids (40-mg prednisone) and several months of topical clobetasol with vitamin D supplementation. She eventually received collagen fillers for the skin indentation with excellent cosmetic results (Fig. 1A). No ocular, oral, or systemic findings were observed.

3.1.2. Case 2

This patient is a right-handed woman who presented at age 34 years to the neurology clinic due to new onset seizures. Her first seizure was a tonic-clonic convulsion while taking a bath. She then began to experience daily staring spells with loss of awareness, lip smacking, and right-hand movements. Left subcutaneous hemifacial atrophy was noted on physical exam and prompted the diagnosis of focal epilepsy due to PRS. MR imaging after her first seizure showed diffuse left hemispheric atrophy with cortical and subcortical T2 hyperintensities. The patient was initially titrated up to 250-mg lamotrigine and 3000-mg levetiracetam with only rare staring spells reported, and MRI at one year of follow-up was stable. However, after two years on this therapy, seizures began to increase in frequency (daily) and severity (both focal and secondary convulsions) with noticeable cognitive decline. A trial of two 1000-mg rituximab infusions provided no improvement in seizure frequency. She was transitioned from lamotrigine to 1800-mg oxcarbazepine because of insomnia, and three years later 10 mg of clobazam was added. On this regimen, she currently experiences zero to four seizures per month. Most recent MRI at five years follow-up showed slowly progressive left hemispheric atrophy and gliosis (Fig. 1Bi and Bii). Interictal EEGs have consistently shown left hemispheric slowing. Video-EEG (VEEG) monitoring showed interictal epileptiform discharges within the left mid-to-posterior temporal lobe. Epilepsy surgery was not considered due to the potential for devastating loss of language function.

The patient recalled first noticing deepening of the left nasolabial fold three years prior to seizure onset with slowly progressive facial atrophy. She also had a six-year history of right-hand weakness that was attributed to resection of a malignant leiomyosar-coma of the right arm. This weakness is now believed to be central in origin given atrophy of her left motor cortex (Fig. 1Bi). Retrospective examination of an MRI performed after a bicycle accident eight years prior to seizure onset revealed subtle signs of left hemispheric atrophy, raising the possibility that the disease onset began with cerebral atrophy up to a decade prior to seizure onset. She had no ocular, oral, or systemic symptoms.

3.1.3. Case 3

This patient is a right-handed woman who came to our clinic at age 22 years. She began experiencing seizures at the age of 17 characterized by debilitating left foot clonus without loss of awareness and drop attacks. Right-sided facial atrophy began at the age of 3, and she was diagnosed with PRS at the age of 9. In addition to skin findings, she had anisocoria and enophthalmos, but no other systemic symptoms. She underwent several facial reconstructive surgeries, and her dermatologic disease is now stable.

Initial MR imaging following seizure onset showed mild atrophy of the right hemisphere and within it multiple T2/FLAIR hyperintensities scattered throughout gray and white matter, with some in the peri-Rolandic region close to the interhemispheric fissure (primary motor cortex, leg area). When initially seen at our hospital five years after seizure onset, her ASM regimen consisted of four daily ASMs: 400-mg topiramate, 400-mg lacosamide, 2700-mg felbamate, and 20-mg clobazam, and immune-modulatory therapy, monthly alternating intravenous immunoglobulin (IVIg) and methylprednisolone. However, she continued to experience daily focal seizures and monthly drop attacks. A vagal nerve stimulator implanted two years after seizure onset did not reduce seizure burden. Over the next decade, other ASMs were tried, including carbamazepine, levetiracetam, lamotrigine, brivaracetam, phenytoin, perampanel, oxcarbazepine, and valproate, all to high dose and without success. Hydroxychloroquine and methotrexate, rituximab, and plasmapheresis were trialed 7 and 10 years after seizure onset, respectively, and were also ineffective. MRI findings were stable for nearly a decade, but at 11 years of follow-up she developed intractable focal status epilepticus (SE) originating from the right posterior region requiring general anesthesia. Imaging at that time showed further atrophy and gliosis on the right (Fig. 1Ci–iii). SE was resistant to anesthetics, cannabidiol, anakinra, and steroids, and she underwent a posterior disconnection surgery complicated by subdural fluid collection and stroke. The patient and parents had always rejected any resective surgery that could lead to motor deficits. Current daily ASM regimen is 583.2-mg phenobarbital, 1600-mg topiramate, and 4200-mg felbamate. She continues to experience up to 10 focal seizures daily and twice monthly drop attacks.

3.2. Systematic literature review

Database searches returned 377 unique articles, of which 193 were excluded after abstract review or because they were not available (Fig. 2). The remaining 184 articles underwent full text review; 99 articles met all inclusion criteria and contained individualized data on 137 individuals with appropriate diagnoses of either PRS, LScs, or localized scleroderma and epilepsy [6,11–108]. In addition to these 137 individuals, the three cases listed above are included in the data set, for a total of 140 individuals affected. Data are presented in aggregate, with only outliers or special cases specifically referenced in Sections 3.2.1–3.2.5.

Fig. 2.

Flowchart of the systematic literature review. Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) protocol flow diagram for cases of Parry–Romberg syndrome (PRS, also known as hemifacial atrophy) or linear scleroderma (LS) with seizures or epilepsy. 99 publications were included in the final analysis with 137 patient cases from prior publications, in addition to the three new cases (Cases 1–3) presented in this manuscript, for a total of 140 total cases.

Of the 99 articles included, 51 (51.5%) were from neurology or neuroscience specialty journals, including pediatric neurology, and contained approximately half of the cases in this analysis (summarized in Fig. 3A as “Neurology”). The other half of the cases were from 48 (48.5%) non-neurology specialty journals including rheumatology, dermatology, and radiology (listed individually in Fig. 3A). Note that all pediatric or medicine journals not specifically connected to a specialty were included under the labels “General Pediatrics” or “General Medicine” (e.g., Pediatrics, European Journal of Pediatrics, BMJ Case Reports, Medwave, etc.).

Fig. 3.

Description of included literature review articles. (A) Journals by specialty. The table shows the number of publications included in the systematic literature review by primary medical specialty and the number of cases associated with each type of publication. The specialties represented in this study were: Neurology, Rheumatology, Dermatology, General Pediatrics, General Medicine, Radiology, Ophthalmology, Medical Genetics, and Psychiatry. All specialty categories which included pediatric-specific publications are within “General Pediatrics”. (B) Analysis for case reporting bias between neurology and non-neurology publications. To analyze the possibility that neurology-specific publications were more likely to report severe or refractory cases and bias the analysis, a comparison of seizure outcomes reported between neurology and non-neurology publications was performed. Neurology publications reported 74 cases with seizure outcomes following treatment, seizure free were 36% (n = 27), improved were 23% (n = 17), refractory 11% (n = 8), and not specified were 30% (n = 22). Non-neurology publications reported 63 cases with seizure outcomes following treatment, seizure free were 41% (n = 26), improved were 16% (n = 10), refractory 11% (n = 7), and not specified were 32% (n = 20). Two-way analysis of variance showed no significant effect by publication type (p>0.9999). A detailed description of seizure outcome and treatment analysis can be found in Section 3.2.5 and Fig. 8.

To evaluate for the possibility of bias in publication of more severe or refractory cases from neurology journals versus non-neurology journals, final seizure response to treatment reported in each case were evaluated between these journal categories (Fig. 3B). Seizure outcomes and treatments are described in detail in Section 3.2.5, and categories of treatment response included seizure free, seizure improvement, and refractory to treatment. Both categories of journals contained similar percentages of cases for each specified seizure outcome.

The majority of articles included in our analysis were isolated case reports or case series, but eight articles were considered to be of higher methodological quality due the performance of unbiased retrospective chart review of a larger cohort of patients with either LScs, PRS, or both, with or without epilepsy, over a specified period of time. These studies were represented in journals of neurology (n = 2), dermatology (n = 2), neuroradiology (n = 2), rheumatology (n = 1), and pediatrics (n = 1) [15,24,29,33,40,63,85,89]. These chart reviews described a combined total of 253 individuals with LScs, PRS, or both, with epilepsy occurring in 33 cases. The estimated prevalence of epilepsy in PRS/LScs from these publications combined was therefore 13.0% (range: 6.3–50.0%; mean: 25.4%).

3.2.1. External Manifestations

Our cases included 81 females and 59 males for a female-to-male ratio of 1.37 to 1 (Fig. 4A). The most common diagnoses were PRS, LScs, or PRS plus LScs (Fig. 4B). Two cases with linear scleroderma of the head that was not explicitly described as “coup de sabre”, were grouped with LScs [38,65]. Four cases had either LScs of the body or linear scleroderma without head involvement (LS-Body) [44,73,76,77]. One patient had both facial LScs and plaque morphea of the chest, and another patient had both facial LScs with bilateral plaque morphea of the elbows [57,59]. The left side of the head or body was more likely to be involved than the right side, with only a few cases describing bilateral involvement or non-lateralizing involvement at the midline (Fig. 4C).

Fig. 4.

Distribution of age, sex, diagnoses, and localization of external disease. (A) Sex distribution. Individuals were reported as female in 58% (n = 81/140) of cases and male in 42% (n = 59/140) of cases, giving a female to male ratio of 1.37. (B) Diagnoses. 46% had Parry–Romberg syndrome (PRS, n = 64/140), 38% (n = 53/140) had linear scleroderma en coup de sabre (LScs), 14% (n = 20/140) had PRS plus LScs, and 4% (n = 6/140) had LS of the body. Note that n = 3 individuals had multiple diagnoses. (C) Laterality of external findings. 56% (n = 79/140) were on the left side of the face/body, 39% (n = 54/140) were on the right side of the body, 4% (n = 6/140) found lesions on both sides of the face or body, and 1% (n = 2/140) were described as located along the midline. Note that n = 1 individual had different laterality described for separate diagnoses of the face and body. (D) Location of external disease findings. 74% scalp (n = 104/140), 71% face (n = 99/140), 24% ocular involvement (n = 34/140), 11% oral involvement (n = 15/140), 5% neck (n = 7/140), 5% upper extremity (n = 7/140), 3% lower extremity (n = 4/140), 4% chest (n = 5/140), 0.7% back (n = 1/140), 1% hemibody (n = 2/140). Note that 66% (n = 93/140) had multi-site involvement, most commonly of both the face and scalp.

Fig. 4D shows in more detail the sub-regions of the head and body affected by external disease. Note that 66% (n = 93/140) of cases had multi-site involvement, in most, both the face and scalp. Ocular involvement included enophthalmos, anisocoria, uveitis, retinitis, and peri-orbital inflammatory changes. Oral involvement included scarring of the palate, tongue atrophy, and changes to dentition. A minority of patients had skin changes over the neck, chest, back, or upper and lower extremities. Two patients with a diagnosis of PRS had severe atrophy of one side of the entire body (i.e., hemibody).

3.2.2. Age of onset

External disease started in 110 cases (79%) before age 18 years (median age 7 years, range 0–79 years); seizures started in 81 patients (58%) before age 18 years (median age 13 years, range 0–80 years). The timing of onset for both external disease manifestations and neurological conditions was reported in 108 cases (Fig. 5). Two-thirds of cases experienced seizures after external onset, while only 19% of cases experienced seizures before external onset. Median time to seizure onset from onset of external disease was 2.75 years (range: [C0]8 years to +36 years; mean: 5.0 ± 7.2 years). In almost a third of cases, seizures occurred within one year of recognition of external manifestations (hair, skin, eye, mouth). To further break down risk surrounding this first year, 6% (n = 7/108) had seizure onset in the months just before external manifestations were recognized, 15% (n = 16/108) had seizures and external disease noted at the same time, and 8% (n = 9/108) had seizure onset in the months just after external disease began. After that first year, risk of developing seizures appeared to decrease the further from onset of external disease manifestations, with most cases (39%, n = 42/108) occurring within 10 years of external onset. Rarely epilepsy occurred earlier than one year prior to external manifestations (Fig. 5).

Fig. 5.

Timing of epilepsy onset compared to external disease onset. Age of external disease onset was subtracted from age of seizure onset to understand the timing of seizures compared to external findings. Median time to seizure onset from onset of external disease was 2.75 years (range −8 years - +36 years; mean 5.0 ± 7.2 years). In terms of distribution of seizure onset, individuals seemed most likely to develop seizures within one year around the onset of the external disease findings (29.6%, n = 32). Within this first year, seizures occurred in 6% (n = 7/108) in the months just prior to, in 15% (n = 16/108) at the same time, and in 8% (n = 9/108) in the months just after the external manifestations. Epilepsy developed after the onset of external disease in a total of 66% (n = 71/108) of cases beginning months to years after the onset of external disease manifestations, with the risk of seizures decreasing the further from external onset. In order of time since external onset starting at ‘1 yr to 5 yrs’ going to ‘36 yrs to 40 yrs’: n = 21, n = 21, n = 11, n = 5, n = 2, n = 1, n = 0, n = 1. Some individuals, 19% (n = 21/108) developed epilepsy before the onset of the external disease manifestations. In order of time before external onset from ‘−6 yrs to −10 yrs’ to ‘−1 yr to −5 yrs’: n = 1, n = 13.

3.2.3. Seizure characteristics and EEG findings

Of all cases examined, 116 gave descriptions of seizure semiology that allowed for classification according to the ILAE[7], though many did not provide much detail, especially when the case was published in a non-neurological journal (Fig. 6A–E). Overall, seizures with loss of awareness (focal unaware, formerly “complex partial”) were experienced by the majority of individuals affected, and progression to bilateral tonic-clonic convulsion (formerly “secondarily generalized”, or “grand mal”) was seen in over half of cases (Fig. 6A). A minority of cases experienced either SE, epilepsia partialis continua (EPC), or seizures without loss of awareness (focal aware, formerly “simple partial”) (Fig. 6A).

Fig. 6.

Seizure descriptions and EEG diagnostic findings in epilepsy due to Parry–Romberg syndrome (PRS) and linear scleroderma en coup de sabre (LScs). (A) Types of seizures. Seizures without loss of awareness (LOA) occurred in 7% of cases, 78% (n = 90/116) had LOA, including 58% (n = 67/116) with bilateral tonic-clonic convulsions. Status epilepticus (SE) occurred in 16% (n = 18/116) of cases and epilepsia partialis continua (EPC) occurred in 6% (n = 7/116) of cases. (B) Semiological features. The most common focal features noted at seizure onset were motor signs in 28% (n = 33/116) and sensory auras in 10% (n = 132116). Other focal features at seizure onset included visual auras (4%, n = 5/116), automatisms (2%, n = 2/116), olfactory auras (2%, n = 2/116), language disturbance (1%, n = 1/116), and déjà vu (1%, n = 1/116). Seizure onset features were not described or unclear in 52% (n = 60/116) of cases. (C) EEG findings. EEGs were abnormal in 80% (n = 72/90) of cases and 20% (n = 18/90) were normal. (D) Nature of EEG abnormalities. Of the abnormal EEGs, 58% (n = 42/72) reported interictal epileptiform discharges, 18% (n = 13/72) had ictal patterns (seizures), while 15% (n = 11/72) reported focal or generalized slowing, and 8% (n = 6/72) did not specify the abnormal findings. (E) Distribution of EEG abnormalities. Of the abnormal EEGs, 65% (n = 47/72) reported the abnormalities localizing to the brain hemisphere ipsilateral to the skin, scalp, ocular, or oral findings. In only 7% (n = 5/72), EEG abnormalities were contralateral to the external findings, 10% (n = 7/72) were bilateral, 6% (n = 4/72) were generalized, and distribution was unspecified in 13% (n = 9/72).

Focal features suggestive of lobe or hemisphere of seizure origin were described in approximately half of cases in sufficient detail. Motor features indicating unilateral frontal lobe involvement were the most common, followed by somatosensory and then visual auras, indicating possible parietal and occipital onset, respectively (Fig. 6B). Temporal lobe semiologies including olfactory auras, automatisms, déjà vu, and language disturbances were mentioned in a small number of cases (Fig. 6B). On EEG, VEEG, or electrocor-ticography, more than half had interictal epileptiform discharges and almost a fifth had seizure patterns captured (Fig. 6C–D). These abnormalities were predominantly found ipsilateral to the skin changes, but a few cases had contralateral, bilateral, or generalized findings (Fig. 6D).

Seven cases in our analysis had a formal diagnosis of Rasmussen’s encephalitis (RE), with 6 out of 7 of these having PRS and 1 having LScs [18,29,55,59,72,87]. We identified an additional 10 cases that had features of RE such as cerebral hemiatrophy with functional decline, hemiparesis, or hemianopsia, but these cases were not formally classified as RE [27,32,44,67,75,82,86,91,93,103]. Three of the confirmed RE cases and two of the suspected RE cases developed EPC.

3.2.4. Imaging findings

Brain CT (n = 65) or MRI results (n = 123) were reported in 136 cases in our analysis, summarized in Figures 6A–D. Of these, the majority of patients had intracranial imaging abnormalities (Fig. 7A). These abnormalities were predominantly ipsilateral (81%) to the side affected by the external disease manifestations, although some cases were affected contralaterally or bilaterally (Fig. 7B).

Fig. 7.

Imaging findings in epilepsy due to Parry–Romberg syndrome (PRS) and linear scleroderma en coup de sabre (LScs). (A) Frequency of imaging abnormalities. Of cases reporting imaging results, 92% (n = 125/136) had either (or both) CT or MRI abnormalities with only 8% (n = 11/136) being read as normal. (B) Location of imaging abnormalities in relationship to external findings. Of the cases reporting abnormalities, 81% (n = 101/125) had those abnormalities ipsilateral to the side of skin, scalp, ocular or oral involvement, and only 5% (n = 6/125) had contralateral brain findings. 13% (n = 16/125) had bilateral intracranial abnormalities, 2% (n = 2/125) did not specify the side of the abnormalities. (C) Abnormal CT findings. The most common abnormal CT findings were calcifications (58%, n = 29/50), regions of hypodensity (19%, n = 8/50) or hyperdensity (4%, n = 2/50), and regions with contrast enhancement (4%, n = 2/50). (D) Abnormal MRI findings. The most common abnormal MRI findings were T2/FLAIR hyperintensities (70%, n = 79/113), atrophy (44%, n = 49/113) with 19% (n = 21/113) being hemiatrophy and 25% (n = 28/113) being focal atrophy. Gadolinium enhancement was described in 14% (n = 15/113) of cases. Note that 53 individuals had both CT and MR imaging.

Of the 125 total abnormal imaging studies (CT, MRI, or both), 40% (n = 51/125) specified multiple discrete foci in cerebral and deep gray matter or infratentorial structures, while 17% (n = 21/125) described diffuse unilateral hemispheric involvement. In studies with focal findings, the frontal lobe was involved in 46% (n = 58/125) of cases, but lesions were described in all other lobes including parietal (30%, n = 38/125), temporal (26%, n = 33/125), occipital (14%, n = 17/125), and insula (3%, n = 4/125) in either the gray or white matter or both. White matter lesions were emphasized in 47% (n = 59/125) of cases. Deep gray matter structure (basal ganglia and thalamus) involvement occurred 10% (n = 13/125) of cases. Rarely, infratentorial lesions were described (5%, n = 6/125) in cerebellum, midbrain, pons, or medulla. Meningeal involvement was described in 6% (n = 8/125) of cases.

The most common abnormality found on CT was calcifications, and, on MRI, T2/FLAIR hyperintensities (Figure C and D). Nearly half of cases (44%) with abnormal MRI had either focal or hemispheric atrophy. Gadolinium and contrast enhancement were noted in a minority of cases. Cerebral angiography was performed in 30 cases with 75% (n = 23/30) having normal or non-specific findings, while 25% (n = 7/30) had a vascular abnormality (cavernoma, n = 1) or vascular caliber changes occurring ipsilaterally (n = 3), contralaterally (n = 1), and bilaterally (n = 2) compared to other brain lesions.

Imaging follow-up across time frames of months to years was reported in 69 cases, with 54% (n = 37/69) reporting lesion progression, 22% (n = 15/69) reporting lesion stabilization, 9% (n = 6/69) of cases reporting brain lesion regression, and 14% (n = 10/69) with variable patterns, and waxing and waning of different lesions. There was no apparent correlation with seizure outcome and imaging changes. For example, of the 37 cases indicating lesion progression, 73% (n = 27/37) reported seizure freedom or improvement with treatment while 27% (n = 10/37) had refractory seizures.

3.2.5. Seizure treatment and outcomes

Treatment for epilepsy in PRS/LScs-spectrum includes ASMs, immunomodulatory therapy (IMT) to target a suspected underlying autoimmune disorder, and surgery in refractory cases, but there have been no randomized, controlled trials on efficacy of therapies for seizure control in these conditions. In our analysis, 98 cases reported seizure treatment outcomes with the general treatment modalities tried (ASM, IMT, or surgery). Of the cases that reported seizure freedom or seizure improvement, 82 reported the specific treatments resulting in those outcomes. These data are summarized in Fig. 8A–C. The most common outcome reported was seizure freedom or reduction in seizure burden, while only a minority reported seizures refractory to any treatment (Fig. 8A). The most common reported treatments were ASM and IMT (Fig. 8B).

Fig. 8.

Seizure treatment and seizure outcomes in epilepsy due to Parry–Romberg syndrome (PRS) and linear scleroderma en coup de sabre (LScs), available data from entire dataset (A–C). (A) Seizure outcomes. Seizure freedom was achieved in 55% (n = 54/98) of individuals, while 29% (n = 28/95) reported seizure improvement either in terms of reduced severity or frequency, and 16% (n = 16/98) reported seizures that showed no response to any form of treatment. (B) Treatment modalities. Anti-seizure medications (ASMs) were trialed in 92% (n = 90/98), 65% (n = 64/98) used one or more type of immunomodulatory therapy (IMT), and 18% (n = 18/98) attempted to control seizures with a surgical intervention. (C) Treatment response. Response to a single ASM was reported in 27% (n = 22/82) of cases (seizure free 21%, n = 17/82; improved 6%, n = 5/82). An additional 28% (n = 23/82) required switching to or adding more than one ASM (seizure free 20%, n = 16/82; improved 9%, n = 7/82). IMT including, but not limited to, oral steroids, methotrexate, and tocilizumab with or without ASM included were reported to help in 30% (n = 25/82) of cases (seizure free 16%, n = 13/82; improved 15%, n = 12/82). A small number cases, 10%, responded well to surgical intervention (n = 8/82) with or without continuation of an ASM or IMT (seizure free 9%, n = 7/82; improved 1%, n = 1/82). Surgical interventions included hemispheric disconnections and resections. The analysis on seizure outcomes, treatment modality, and treatment response was repeated for the eight higher methodological quality publications, data available for 19 of total of 36 patients, see Section 3.2.5 (D–F). (D) Seizure outcomes. Seizure freedom was achieved in 53% (n = 10/19) of individuals, while 16% (n = 3/19) reported seizure improvement either in terms of reduced severity or frequency, and 32% (n = 6/19) reported seizures that showed no response to any form of treatment. (E) Treatment modalities. Anti-seizure medications (ASMs) were trialed in 84% (n = 16/19), 74% (n = 14/19) used one or more type of immunomodulatory therapy (IMT), and 26% (n = 5/19) attempted to control seizures with a surgical intervention. (F) Treatment response. Positive response to a single ASM was reported in 10% (n = 1/10) of cases. An additional 30% (n = 3/10) required switching to or adding more than one ASM (seizure free 20%, n = 2/10; improved 10%, n = 1/10). IMT with or without ASM included were reported to help in 40% (n = 4/10) of cases (seizure free 30%, n = 3/10; improved 10%, n = 1/10). Frontal and temporal surgical resection resulted in seizure freedom in an additional 20% (n = 2/10), with or without continuation of an ASM or IMT.

Separating seizure outcome by treatment type showed that seizures freedom was achieved in 41% with one or more ASMs (Fig. 8C). The ASMs used when seizure reduction or freedom was achieved included: carbamazepine (n = 22), levetiracetam (n = 14), valproate (n = 9), phenobarbital (n = 8), phenytoin (n = 8), oxcarbazepine (n = 7), topiramate (n = 6), clobazam (n = 6), lamotrigine (n = 5), lacosamide (n = 3), vigabatrin (n = 2), nitrazepam (n = 1), and pregabalin (n = 1). Of the cases responding best to ASMs, 21 had also tried systemic immunotherapy with little improvement beyond that achieved by ASMs. IMTs that resulted in fewer seizures or complete seizure freedom included: systemic corticosteroids (n = 18), often combined with either methotrexate (n = 11), IVIg (n = 2), rituximab (n = 2), mycophenolate mofetil (n = 1), tocilizumab (n = 1), cyclophosphamide (1), or azathioprine (1). Plasmapheresis was noted to cause improvements in seizures in one additional case. Successful surgical interventions were functional hemispherectomy (n = 4) or focal resection (n = 4). However, 5% (n = 5/98) of cases in the refractory group continued to have seizures despite multiple surgical interventions including focal resection (n = 3), hemispherectomy (n = 1), callosotomy (n = 2), and vagal nerve stimulator implantation (n = 3). There was no significant relationship found between the success of any class of treatment (ASM, IMT, surgery) and the timing of seizure onset compared to external disease manifestations (data not shown).

Publications with higher quality methodology and larger patient cohorts were used for a separate analysis to address possible bias in publication of seizure outcomes and treatments compared to case reports, and this analysis was evaluated against the entire set of articles. There were 8 such higher methodological quality publications which contained 33 patients, plus our three cases [15,24,29,33,40,41,48,63,85,89]. Of these 36 cases, 19 reported seizure outcomes and treatment modalities tried. Of the cases that reported seizure freedom or seizure improvement, 10 reported the specific treatments resulting in those outcomes. These data are presented and discussed in Fig. 8D–F. In these cases, proportionally more refractory seizures were reported (32%, n = 6). Response to >1 ASM (30%, n = 3) and IMT (40%, n = 4) were similar to that found for the whole dataset represented in Fig. 8A–C, although more of the seizure free and seizure improved cases responded to surgery (20%, n = 2) and fewer responded to a single ASM (10%, n = 1).

3.2.6. Brain histology

Brain biopsy or tissue resection results were available for 23 cases. The majority of histological descriptions, 74% (n = 17), noted tissue injury such as gliosis, neuron loss, and atrophy (data not shown). Brain inflammation was described in 57% (n = 13) of cases, with 85% (n = 11) of that inflammation being perivascular and 15% (n = 2) not specifying localization. Inflammatory cell type was described as lymphocytic in 85% (n = 11) of those cases. Other vascular abnormalities, such as blood vessel thickening, dilation, or proliferation was described in 39% (n = 9) of cases.

4. Discussion

We hope that these data compiled from our institutional cases and literature review will help neurologists encountering individuals with PRS/LScs-related epilepsy to make more informed decisions regarding diagnostic testing, treatment planning, patient education, and long-term monitoring for the improvement of patient outcomes. In addition to discussing the implications of this study, we will also explore gaps in knowledge that could be addressed by future clinical and basic research.

4.1. Comparison to prior studies

We found our analysis of cases congruent with existing data on PRS and LScs. Previous reports have estimated the prevalence of PRS and LScs to be 0.7–8 in 100,000 [1,109] and 0.4–2.7 in 100,000 [110], respectively, with a female-to-male ratio of 2–3:1 [1]. Prevalence of these conditions within our institution’s catchment area is on the lower end of those estimates for these conditions. In both our cases and the literature review, there was female predominance for PRS/LScs and related epilepsy, suggesting a lack of sex-specific risk factors for developing seizures. Epilepsy has been shown to occur in between 10% and 73% of individuals with PRS or LScs [6,8,24,111,112]. The prevalence in both our case series and literature review fell within this wide range and were very similar to each other, at 16% and 13%, respectively. PRS and LScs may co-occur in up to 20–37% of cases [6,113]. Although there were no cases in our population, 15% of literature review cases had this co-diagnosis.

Previous reviews on neurological sequelae in PRS and LScs have broadly discussed epidemiology, imaging findings, and association with related conditions such as RE, but there is limited population-level information specifically related to epilepsy in these conditions. CNS symptoms are highly variable and have been reported in 17–73% of affected individuals [6,111–113]. The average time from onset of external manifestations to development of any neurological disease (headache, seizure, focal symptoms, neuropsychiatric symptoms) was 4.3 ± 7.3 years, with only 16% of individuals having neurological disease prior to onset of external manifestations [6], similar to our findings for epilepsy. Our study complements a recent literature review by Rocha et al which presented 40 pediatric PRS cases with epilepsy [114] – –from which we included 39 cases that were available to us for the sake of comprehensiveness, in addition to 98 more cases – including LScs patients with epilepsy – discovered by our literature search. With this expanded analysis, we find that a majority of individuals appear to have focal-onset seizures with progression to bilateral tonic-clonic convulsion. In line with previous studies, we find that EEG findings are common and typically ipsilateral to external disease, similar to abnormal brain imaging findings which are seen in over 90% of individuals with PRS and LScs and epilepsy [114]. This is much higher than the rate of brain lesions in individuals with PRS and LScs as a whole (16–44%) [24,33]. Interestingly, compared to the exclusively pediatric cases in the previous review which had 38% refractory cases, in our analysis, which included adults this was far less common with only 16% of a mixed (pediatric and adult) population with seizures unresponsive to any treatment [114]. These data may indicate improved prognosis with age of seizure onset or better responsiveness of patients with LScs. Alternatively, an autoimmune disease may undergo senescence with seizures becoming more drug-responsive over time.

4.2. Considerations regarding diagnosis and testing

Early diagnosis of PRS or LScs in individuals with epilepsy is critical for making timely treatment decisions, and skin findings ipsilateral to imaging abnormalities will be the most important clue to diagnosing these conditions. Neurologists should keep skin or hair changes on the review of systems and checklist for physical examination of persons with new onset epilepsy, especially in the context of a personal and or family history autoimmune disease. In cases of LScs, patients may have a history of cosmetic concerns. Some cases of adult onset PRS have slow progression, and patients and families may be unaware of any changes unless prompted to compare older with recent photographs.

Persons presenting with PRS/LScs, with or without epilepsy, should receive imaging to identify lesions corresponding to neurological signs and symptoms. MRI is ideal to identify atrophy and gliosis. Calcifications and bony changes affecting the hemicranium and facial skull are best seen with CT. Abnormalities are typically non-enhancing, but contrast and gadolinium were useful in a minority of cases to identify areas of active inflammation. Cerebral angiography was commonly reported as normal or non-specific, suggesting vascular pathology may not be a significant driver of the disease. Providers may wish to postpone angiography in these individuals unless specific vascular pathology is on the differential. VEEG should be utilized when surgical interventions for seizures are being considered. Brain biopsies are non-specific and are probably not helpful unless neoplasm is suspected. For all testing, serial examinations can provide information on the natural history of the conditions but would not be necessary in persons with controlled seizures. Cerebrospinal fluid analysis may reveal signs of inflammation, such as oligoclonal bands, but no specific biomarkers for PRS or LScs are known [1].

4.3. Considerations regarding disease mechanisms

Many questions remain about the pathogenesis and chronicity of PRS and LScs, and this study was not designed to disprove or support any specific disease mechanism. PRS and LScs are currently thought to be autoimmune in nature due to high comorbidity with other autoimmune conditions such as hypothyroidism and vitiligo. Tissue (skin, brain) biopsies frequently reveal perivascular lymphocytic inflammation, and the external manifestations (atrophy, scarring, discoloration) can stabilize, or even reverse, with systemic immunotherapy [1,2,5,8,24,115]. Indeed, in our analysis, perivascular lymphocytic infiltration was the most commonly reported inflammatory finding in brain biopsies. Unlike other scleroderma spectrum conditions, however, there is no antibody target yet associated with PRS or LScs. The vast majority of PRS/LScs cases appear sporadically without an inheritance pattern, and there is no identifiable racial predisposition [1]. Environmental triggers have been evaluated by previous studies, such as local skin trauma, although only 12% of 205 PRS patients in a global survey recalled an injury associated with onset of skin atrophy [111]. No viral or bacterial infections have been found to be temporally related or present within tissue samples [1,2]. There is no link to a specific environmental exposure and the fact that this disease has been documented for hundreds, and potentially thousands, of years does not support causation by any specific drug or toxin [3,4]. Ablation of the cervical sympathetic ganglion in animal models results in unilateral facial atrophy resembling PRS [116], but most human cases examined have normal sympathetic nervous system function on the affected side, and thus this cannot explain the skin and CNS findings in all cases [1,2,6,112].

It is likely that PRS/LScs-spectrum disease represents a complex interplay between genetic predisposition to autoimmune disease, developmental processes, and environmental exposures. Skin and nervous tissues derive from the ectodermal germ layer during embryonic development. Multi-potent neural crest cells, a specialized cell lineage from the ectoderm, give rise to the facial skin, cranial ganglia, craniofacial bones, teeth, retina, and brain vascular pericytes. Importantly, LScs lesions are distributed along the lines of Blaschko, which represent the pathways of neural crest cell migration and division during development of the epidermis [117]. To speculate on one possible mechanism, somatic mutations occurring in a lineage of ectodermal or neural crest cells could pre-dispose those cells to immune attack in a genetic background of autoimmune susceptibility. Any localized injury, infection, or toxic exposure to neural, vascular, or epidermal tissues could trigger inflammation that results in immune recognition of and reaction to the mutant cell lineage as foreign. This hypothesis could explain the pattern of disease involvement and the various tissues affected. There is evidence that somatic mutations (also called somatic variants or mosaicism) are one genetic driver of other very rare immune and autoimmune conditions [118,119]. Pursuit of the autoimmune hypothesis linked to somatic mutation could be advanced by examination of epithelial cells and neurons with genome sequencing and transcriptome analysis in the PRS/LScs population.

Epilepsy in these conditions likely arises as a non-specific result of focal chronic inflammation or later gliotic scar formation. Expanding or recurrent inflammation and gliosis in other foci, as well as kindling by frequent seizures may establish new epileptogenic circuitry over the course of years. Again, while this review was not designed to investigate mechanisms, the high likelihood of epilepsy onset within the first year of detection of cutaneous disease suggests that common inflammatory mechanism of skin and brain disease may be a major engine of seizures in this population. This may have implications for approach to treatment.

4.4. Considerations regarding treatment and outcomes

The three new cases presented here showcase the spectrum of possible treatment outcomes for PRS/LScs-related epilepsy. Encouragingly, our literature analysis found that over half of people with PRS/LScs-related epilepsy can achieve seizure freedom with available treatments: ASM, IMT, and surgery.

Guidelines for treatment in PRS/LScs-related epilepsy could help with understanding of which therapy – particularly ASM and IMT, at what timepoint and in which patients, is most helpful for seizure control. We propose two possible initial courses of action: 1 – try one or two ASMs to maximal dose, then escalate to ASM with IMT or 2 – try both an ASM and IMT combined. Regardless of the choice in therapy for seizures, early consultation with a rheumatologist or dermatologist is recommended, if not already involved, as IMT may benefit the external manifestations of disease in addition to seizure control. Corticosteroids and methotrexate were the most commonly used with success in this analysis; however, IMT selection should be based on side-effect profile and cost for the individual patient. If these therapies fail to control seizures, surgical intervention could then be discussed, though any lesion removal is accompanied by the worry that another lesion could arise in a different location, and a resective approach may not be successful unless a hemispherectomy or hemispherotomy is done.

Conceptually, epilepsy related to autoimmune processes early in the disease due to acute and chronic inflammation, and in later years associated gliosis, may lead us to hypothesize that ASMs in combination with IMTs would be most effective early in the course of the disorder by targeting both neuronal excitation and inflammation. Later, when chronic inflammation has resulted in tissue injury and scar formation, we would hypothesize that ASMs and surgical management are more effective against seizures. While our analysis of onset versus treatment showed a trend in support of this hypothesis, the data were not statistically significant.

There are currently no validated biomarkers to track inflammation in this condition that could give clues as to the optimal timing of different type of treatments. Discovery of such biomarkers, e.g. through analysis of inflammatory markers in the CSF, could provide invaluable tools for future treatment recommendations, including predicting the likelihood of success of IMT versus using ASMs alone.

4.5. Advocacy for a database and consortium

The future of understanding PRS/LScs-related epilepsy and best treatments, particularly the benefit of ASM versus IMT, lies in improved case tracking and research. The rarity of this condition requires multi-national, multi-site collaboration for capture and analysis if any progress is to be made. As a start to such an endeavor, all literature review data have been deposited in a publicly available database at Synapse.org at http://www.synapse.org/EpilepsyinPRSandLSCS. We kindly ask any individuals with access to additional published case reports to upload them to this database. Our goal was to create an international database where providers can share and access de-identified data for the purpose of tracking treatments and outcomes in this patient population. Cases contributed to the patient database by new research teams must have home institution IRB approval for collection and analysis, but once de-identified for upload, this is not considered personal health information and can be used or disclosed without limitation. Instructions on navigating the database is detailed in the Supplemental Fig. 2, and information on contributing can be found on the Synapse.org database Wiki page. This database could also be used for connecting researchers interested in collaborating on this endeavor.

4.6. Limitations

This paper presents Class IV evidence and has potential for bias. Concerns were that (1) reporting in neurological journals would be more likely to yield cases with neurological complications, including epilepsy, compared to publications in non-neurological journals; and (2) case reports would highlight more severe pathology compared to articles with larger patient cohorts. We found no evidence that neurology versus non-neurology journals were skewed toward publication of more severe epilepsy and worse seizure outcomes, suggesting that the data from all types of publications can be weighted equally in our analysis. In addition, separate analysis of the publications with higher methodological quality showed a similar pattern of results, suggesting validity of the analysis of the entire dataset. Nevertheless, the main limitations of this review are related to the retrospective nature of the new cases presented and the non-standardized collection of data points in the published literature. The cases published may not be representative of the full range of disease presentations and may only constitute a snapshot of the affected individuals in time. Any trends or patterns found could be due to chance or bias related to publishing cases with positive findings.

Abbreviations:

ASM

anti-seizure medication

CARE

case reporting guidelines

CNS

central nervous system

CSF

cerebrospinal fluid

CT

computed tomography

EEG

electroencephalography

EPC

epilepsia partialis continua

F

female

GTC

generalized tonic-clonic

ILAE

International League Against Epilepsy

IMT

immunomodulatory therapy

IRB

institutional review board

IVIg

intravenous immunoglobulin

LS

linear scleroderma

LScs

linear scleroderma en coup de sabre

M

male

mg

milligram

MRI

magnetic resonance imaging

PHA

progressive hemifacial atrophy

PRISMA

preferred reporting items for systematic reviews and meta-analyses

PRS

Parry–Romberg syndrome

RE

Rasmussen’s encephalitis

SE

status epilepticus

T2/FLAIR

T2 weighted imaging with fluid attenuated inversion recovery

UCH

University of Colorado Hospital

VEEG

video-EEG