Abstract
Objective:
Parry-Romberg syndrome (PRS) and en coup de sabre (ECDS) are subtypes of craniofacial localized scleroderma. Systematic analyses of central nervous system imaging findings and their clinical associations in children are lacking. Here, we aim to characterize neuroimaging findings and associated neurological symptoms in these conditions.
Methods:
Neuroimaging and neurological symptoms of children evaluated at our institution with a diagnosis of PRS or ECDS were retrospectively reviewed. Laterality, location, stability, and number of lesion(s) were evaluated, as was the presence of susceptibility lesion(s) and contrast enhancement. History of seizures or headaches was noted.
Results:
From 2003 to 2019, 80 patients with PRS or ECDS were followed at our institution. Neuroimaging was completed in 73 and found to be abnormal in 25. In 12 (48%) of these 25 cases, headaches and/or seizures were present. In the vast majority of these cases (22/25, 88%), lesions were ipsilateral to skin findings. White matter was involved in 19 (76%) patients. MRI abnormalities preceded a rheumatological diagnosis in 7 (28%). Susceptibility lesions were noted in 11 (44%), and 8 (73%) of these patients endorsed a history of headaches. Most lesions were in the supratentorial compartment, did not enhance, and were stable at 1-year follow up imaging. Of those with progression, susceptibility findings were present at baseline.
Conclusions:
Neuroimaging findings in pediatric PRS and ECDS are often supratentorial, stable, unilateral, and ipsilateral to skin findings, and they can precede cutaneous findings.
Parry-Romberg syndrome (PRS) and en coup de sabre (ECDS) are two forms of craniofacial localized scleroderma grouped in the linear scleroderma of the head subtype, with a mixture of both superficial cutaneous findings and deeper tissue findings that may lead to hemi-atrophy of the face. Given these conditions can progress and are thought to be autoimmune with inflammatory-driven fibrosis and atrophy, rheumatologists often employ steroids or disease-modifying therapies to halt progression. Increasing evidence supports the notion that the central nervous system is often involved in PRS and ECDS. In a recent large single-center retrospective review, Doolittle and colleagues found that nearly half of the neuroimaging studies completed in their patients with PRS or ECDS were abnormal. 1 Moreover, the most common neurological symptoms reported in this cohort were seizures and headaches, and nearly one-third of the patients with neuroimaging abnormalities were neurologically asymptomatic.
Studies of central nervous system involvement in pediatric PRS and ECDS are largely limited to case reports and case series. 2 -5 Chiu and colleagues 6 described 19 children with PRS or ECDS who had brain magnetic resonance imaging (MRI) performed. Intraparenchymal abnormalities were appreciated in 4 (21%), all of which consisted of T2-weighted hyperintensities. By reviewing the literature, the same study identified 51 published children with a diagnosis of PRS or ECDS who had some form of neuroimaging reported, with the majority showing an intracranial abnormality on either head computed tomography (CT) or brain MRI. To obtain a more accurate estimate of the rate of neuroimaging abnormalities in PRS/ECDS, we systematically reviewed all children consecutively diagnosed with PRS or ECDS at our institution over a 16-year span.
Methods
Patients
A retrospective chart review was performed utilizing an Institutional Review Board-approved clinical database at the University of Pittsburgh, the National Registry of Childhood Onset Scleroderma (PRO #14080297). All patients with a diagnosis of PRS and/or ECDS made between 6 months and 18 years of age that were at seen UPMC Children’s Hospital of Pittsburgh and enrolled in the National Registry of Childhood Onset Scleroderma between January 1, 2003, and December 31, 2019 were identified (Figure 1). For the purposes of this study, PRS was defined as unilateral global undergrowth or atrophy of the soft tissue, muscle, and/or osteocartilagenous structures of the face with minimal or absent cutaneous features. ECDS was defined as unilateral linear band(s) of fibrotic and/or atrophic cutaneous and subcutaneous tissue of the scalp and/or face, variably involving the muscles and osteocartilagenous structures of the same geographic plane and associated with prominent superficial features (see examples in Figure 2). Among those patients with a diagnosis of PRS or ECDS, children who had head CT or brain MRI completed were included in the study.
Patient selection flow diagram. Of the 545 pediatric patients with scleroderma that are part of the National Registry of Childhood-Onset Scleroderma database, 80 had craniofacial scleroderma (PRS and/or ECDS), and 73 had some form of neuroimaging performed. In 25 of these children, brain MRI was abnormal.
Representative spectrum of clinical findings in craniofacial scleroderma PRS and ECDS patients with abnormal neuroimaging. (A) Patient with two right-sided ECDS lesions (hyperpigmentation with dermal and subcutaneous atrophy in a band-like distribution; arrows) of the forehead extending to the superior orbital rim. (B) Patient with right-sided lower face ECDS lesion (hyperpigmentation with dermal and subcutaneous atrophy in a band-like distribution; arrow) over mandible extending through the inferior vermilion border and left-sided PRS involvement consisting of hemifacial atrophy of the zygomatic region (bracket). (C) Patient with left-sided PRS involvement consisting of hemifacial atrophy of the infraorbital, zygomatic and mandibular areas (bracket). Note the absence of significant cutaneous changes of affected area.
Review of Clinical Documentation
Database and electronic medical records of each patient were reviewed in detail. The following data were systematically extracted from the National Registry of Childhood Onset Scleroderma database and supplemented by review of pediatric rheumatology and child neurology clinical notes: date of PRS or ECDS diagnosis, type(s) of localized scleroderma present, laterality of localized scleroderma, neurological symptoms, neurological examination abnormalities, and available electroencephalogram findings.
Neuroimaging Analysis
Neuroimaging was analyzed independently by 2 child neurologists and 1 pediatric neuroradiologist. All available head CT, brain MRI, and brain magnetic resonance angiography were reviewed. Acquisition dates of all scans were documented. These dates were compared to dates of PRS or ECDS diagnosis to denote patients who demonstrated neuroimaging abnormalities prior to a rheumatological diagnosis.
All acquired images for all patients were then systematically analyzed. For those patients who had a head CT performed, presence of at least 1 hyperdensity or hypodensity was recorded. For brain MRI studies, the following data were extracted: involvement of the gray matter, involvement of the white matter, involvement of the posterior fossa, presence of 1 or more susceptibility lesions, and enhancement of 1 or more lesion with gadolinium. In patients with T2-weighted abnormalities of the white matter, lesions were further classified based on the presence of 1 singular well circumscribed lesion, presence of multiple lesions, or presence of at least 1 diffuse poorly circumscribed lesion. In the subgroup of patients who had brain magnetic resonance angiography performed, the presence of 1 or more abnormalities in vessel structure was documented. Finally, for those patients who had repeat imaging (defined as a second scan of the same modality performed at least 1 year following the first scan in which an abnormality was noted), lesion stability was noted. In all instances, criteria were classified as present or not present.
The laterality of all neuroimaging findings in our patients was documented. For the patients whose neuroimaging findings were unilateral, the laterality of the radiological findings was compared to the laterality of dermatological findings. Patients whose neuroimaging findings were bilateral were included in a separate group for analysis. Given multiple lesions were present in all patients whose neuroimaging findings were bilateral, lesion number within each hemisphere was quantified to determine laterality of greater lesion occurrence.
Statistical Analyses
Continuous data are presented as median (interquartile range). Categorical data are presented as number (percentage). Categorical data were compared using 2-sided Fisher’s exact tests. All statistical analyses were performed with Stata, version 15 (StataCorp LP, College Station, TX).
Results
Abnormal Neuroradiological Findings Are Frequently Appreciated in Pediatric PRS and ECDS
A total of 80 children were diagnosed with PRS and/or ECDS between January 1, 2003, and December 31, 2019 (Figure 1). Forty-seven (59%) were female. At least 1 form of neuroimaging was completed in 73 (91%) (Table 1). In all cases, brain MRI was available. In 10 (13%) cases, head CT was also performed, and in 13 (16%) cases, some form of angiography (CT angiogram or magnetic resonance angiography) was available for review. Of these 73 patients, 25 (34%) had at least 1 abnormality appreciated on neuroimaging and 16 of these (64%) were female. In the 48 patients with normal brain imaging, repeat serial imaging was performed in 10 children. The median age at the time of PRS and/or ECDS diagnosis in the 73 patients with imaging was 8.5 years (interquartile range [IQR] 5.6–11.4), while the 25 children with abnormal imaging were slightly younger (median 8.2 years, IQR 5.7–9.9). The median age at the time of the first abnormal neuroradiological exam was 7.4 years (IQR 6.2–9.8). Notably, in 3 (12%) of these patients who ultimately had abnormal neuroimaging findings, the first neuroimaging exam performed was normal.
Characteristics of Craniofacial Scleroderma PRS or ECDS Children Who Had Neuroimaging Completed.
Abbreviations: ECDS, en coup de sabre; IQR, interquartile range; NA, not applicable; PRS, Parry-Romberg syndrome.
In general, ECDS was more common than PRS, and the distribution of PRS and ECDS was similar between those with and without abnormal brain imaging. Specifically, in the 48 patients with normal imaging, 6 (13%) had PRS only, 27 (56%) had ECDS only, and 14 (29%) had both PRS and ECDS. Dermatological findings were present on the right side in 20 children (42%), the left side in 27 (56%), and bilaterally in 1 (2%). In the patients with abnormal neuroimaging, 5 (20%) had PRS only, 12 (48%) had ECDS only, and 8 (32%) had both PRS and ECDS. PRS or ECDS findings was present on the right side in 7 children (28%), the left side in 17 (68%), and bilaterally in 1 (4%). In 22 patients (88%), MRI abnormalities were present on the same side as dermatological findings.
The White Matter Is the Most Commonly Affected Central Nervous System Structure in Pediatric PRS and ECDS
Of all neuroimaging modalities, brain MRI was the most sensitive test to screen for neuroradiological findings in children with PRS or ECDS. Abnormalities on brain MRI were appreciated in 25 patients (Table 2). A head CT was performed in 8 (32%) of the 25 cases; hypodensities and/or hyperdensities were found in 4, while the other 4 did not demonstrate any abnormalities, despite their abnormal findings on brain MRIs.
Neuroradiological Findings in a Cohort of Children with PRS and/or ECDS.
Abbreviations: CT, computed tomography; ECDS, en coup de sabre; L, left; MRI, magnetic resonance imaging; NA, not available; PRS, Parry-Romberg syndrome; R, right; WM, white matter.
a In patients with bilateral imaging findings, laterality to skin findings was determined by using the side with greater number of lesions on imaging.
On review of all brain MRI sequences, the most common lesion consisted of white matter hyperintensities in T2-weighted sequences (Figure 3), which were present in 19 (76%) of the 25 cases. In 1 (4%) case, abnormalities were restricted to the gray matter. Additionally, in 7 (28%) additional cases, both gray and white matter abnormalities were present. In nearly one-third of the cases (8 of 25; 32%), radiological abnormalities were multifocal. In 6 children (24%), a large, poorly circumscribed lesion was present. Lesions were restricted to a single hemisphere in 22 (88%) of the 25 cases, while the others were bilateral. In most cases, lesions were restricted to brain regions above the tentorium cerebelli, as only 1 patient had a lesion involving the posterior fossa (eg, pontine hyperintensity and ultimately pontocerebellar atrophy). Surprisingly, 1 or more susceptibility lesions were present in 11 of the 25 cases (44%) (Figure 4).
Hyperintensities on T2-weighted sequences were the most common neuroimaging finding in a cohort of children with PRS and ECDS. Axial T2-weighted fluid-attention inversion recovery sequences of multiple patients with unifocal, multifocal, and diffuse hyperintensities are shown. Of note, panels A-E are from the same patients whose susceptibility-weighted lesions are shown in panels A-E of Figure 4. Additionally, subcategories used for analysis in the manuscript are included in the parentheses below. A: patient 5 (multifocal), B: patient 21(unifocal), C: patient 16 (multifocal and diffuse), D: patient 6 (multifocal and diffuse), E: patient 23 (diffuse), F: patient 25 (multifocal), G: patient 15 (multifocal and diffuse), H: patient 14 (multifocal and diffuse).
Lesions on susceptibility-weighted imaging sequences were the second most frequently encountered neuroimaging finding in children with PRS and ECDS. This increased susceptibility on susceptibility-weighted imaging sequences was present in gray matter and/or white matter and was often punctate or linear. Most children with susceptibility lesions found on their brain MRI endorsed a history of headaches. Of note, panels A-E are from the same patients whose T2-weighted lesions are shown in panels A-E of Figure 3. A: patient 5, B: patient 21, C: patient 16, D: patient 6, E: patient 23, F: patient 4, G: patient 12, H: patient 19.
Gadolinium contrast was administered in 17 (68%) of the 25 children. Postcontrast enhancement was present in only 1 of these cases and consisted of enhancement of the leptomeninges, internal capsule, occipital lobe, and basal ganglia, all ipsilateral to the patient’s dermatological findings. Magnetic resonance angiographies were performed in 14 (56%) of the 25 cases and were abnormal in only 2 patients, consisting of prominent vessels ipsilateral to dermatological findings in 1 patient and a diminutive posterior communicating artery infundibulum ipsilateral to the dermatological findings in another.
In 18 (72%) of the 25 cases, repeat neuroimaging was performed at least 1 year following the first abnormal scan to assess for stability. In 16 of these 19 cases (84%), all lesions were stable, while neuroradiological progression was appreciated in the other 3 (16%). In the 3 cases that showed progression, imaging showed involvement of both the gray and white matter as well as the presence of lesions on susceptibility-weighted images. In the first case, radiological progression was not accompanied by any clinical worsening, and neuroradiological findings included progressive T2-weighted hyperintensities as well as occipital lobe enhancement. This patient continued to show gadolinium enhancement and neuroradiological progression across 7 scans until stability was achieved 3.7 years after his first abnormal MRI. In the second patient, radiological progression consisting of a new susceptibility lesions in the thalamus was associated worsening headaches. In the third child, larger T2-weighted hyperintensities of corona radiata and thalamus were noted at a time when she developed increasing pigmentation and atrophy of her cutaneous lesion. In all 3 cases, the progressing neuroradiological features were ipsilateral to the patients’ dermatological findings.
Headaches and Seizures Occur in Children With PRS/ECDS
Thirty-five (48%) of all 73 children with craniofacial scleroderma endorsed a history of headaches. Across all patients with headaches, regardless of whether neuroradiological abnormalities were present, lesions on susceptibility-weighted imaging were present in 8 (23%) of 35 cases. In contrast, susceptibility lesions were present in only 3 (8%) of the 38 children without headaches. This difference was not significantly different (P = .10). In the subgroup of patients with abnormal neuroimaging, history of headaches was endorsed by 10 children, and 8 (80%) of these children also had susceptibility lesions appreciated on their MRIs. In all cases, these lesions on susceptibility-weighted images were ipsilateral to the patients’ dermatological findings. Diffusion restriction was not present in any of our patients, including those cases with neuroradiological progression.
A history of seizures was present in only 5 children with abnormal MRIs and in 0 children with normal MRIs. An abnormality of the gray matter was appreciated in only 1 child. However, susceptibility lesions were present in 4 (80%) of the 5 children with a history of seizures, while they were present in only 6 (9%) of the 68 children without seizures, a difference that was statistically significant (P = .001). EEG abnormalities were noted in all 5 patients: 1 patient had generalized discharges only, 2 patients had focal discharges and focal slowing, and 2 patients had focal slowing only. Neurological exams were abnormal in only 2 patients, consisting of hemiparesis in 1 patient and dysmetria and ataxia in the other.
An MRI abnormality was discovered before a rheumatological diagnosis in 7 (28%) children. Four (57%) of these 7 patients had neuroimaging performed after presenting with seizures, while 1 had imaging performed because of headaches, 1 had imaging performed because of prematurity, and 1 had imaging performed because of precocious puberty. In these 7 patients, the median time between the first abnormal scan and the date of rheumatological diagnosis was 1.6 years (IQR 0.6–2.4). In 2 of these children, dermatological abnormalities were appreciated at the time of this imaging, leading to prompt rheumatological diagnosis. In the other 5, however, no dermatological abnormalities were initially present, and the median time between imaging and rheumatological diagnosis was 2.3 years (IQR 1.6–2.4).
Discussion
To our knowledge, this is the largest report of neuroimaging abnormalities in a group of pediatric patients consecutively diagnosed with PRS and/or ECDS in a single center. The results of this retrospective study indicate that neuroimaging findings are appreciated in over a third of children with PRS and ECDS. Hyperintensities on T2-weighted sequences and/or lesions on susceptibility-weighted images were the 2 most frequently appreciated neuroimaging abnormalities. Most often, neuroradiological findings were supratentorial, unilateral, stable, nonenhancing, and ipsilateral to the dermatological findings of the patients’ localized scleroderma. As was appreciated in 7 of the patients with abnormal neuroimaging, neuroradiological abnormalities can precede the rheumatological diagnosis of PRS and ECDS, undoubtedly contributing to a diagnostic dilemma for the neurologist.
In nearly three-quarters of our pediatric cohort with abnormal neuroimaging, 1 or more white matter hyperintensity on T2-weighted sequences was present. Multiple prior studies in PRS and ECDS have similarly reported a high occurrence of T2-weighted hyperintensities. Chiu and colleagues 6 reported a series of 4 pediatric patients with PRS and/or ECDS with abnormal brain MRI. In all 4 cases, the MRI abnormalities consisted of T2-weighted hyperintensities of the white matter. It is notable that the age of diagnosis of these children were slightly younger than our patients with T2-weighted hyperintensities (mean ± standard deviation: 5.0 ± 2.5 vs 8.1 ± 3.4 years). In Doolittle and colleagues’ 1 larger cohort of older patients with abnormal neuroimaging, the frequency of T2-weighted hyperintensities (14 of 19, 74%) was similar to what we observed (18 of 25, 72%). Maloney and colleagues 5 also reported that T2-weighted hyperintensities were the most common neuroimaging finding in their cohort of craniofacial scleroderma patients. Neuropathological specimens of T2-weighted hyperintensities in patients with PRS and/or ECDS have shown perivascular T-cell lymphocytic infiltrate and dilated vessels with intimal thickening and calcifications. 7 -10 While the underlying pathophysiology of PRS and ECDS remains unclear, these pathological observations support the notion that an early neuro-inflammatory response may accompany the gliosis occurring with most common neuroradiological findings of patients with PRS and/or ECDS.
Perhaps more surprising was our result that 11 of the 73 children (15%) with PRS and/or ECDS had findings on susceptibility-weighted imaging. Gradient-echo susceptibility-weighted imaging is a relatively new MRI technique that allows for enhanced visualization of blood products, because it emphasizes the paramagnetic properties of hemosiderin, deoxyhemoglobin, and intracellular methemoglobin. 11 Hemosiderin lesions were reported in 2 patients in Doolittle et al’s cohort, 1 and Maloney et al 5 describe susceptibility lesions in many of their patients with PRS and ECDS. Reassuringly, repeat imaging in nearly all children in our cohort with lesions on susceptibility-weighted images was stable without any evidence of progressive intracranial hemorrhage. Given that systemic vasculopathy with thickened neointima and abnormal endothelial cells is frequently appreciated in localized scleroderma, 12,13 it is possible that these susceptibility lesions are associated with a cerebral vasculopathy.
Magnetic resonance angiography did not appear to be a sensitive measure to further characterize these lesions on susceptibility-weighted images, as most of these angiograms were unremarkable in our cohort of patients. Thus, defining the precise vascular phenotype in this subset of patient may require more invasive and clinically impractical tools, such as neuropathological specimens or digital subtraction angiography. The application of digital subtraction angiography to PRS and ECDS has been limited to case reports with variable findings. 14 -16 Of note, digital subtraction angiography was performed in 1 of our patients (patient 6) and was normal. Further studies are needed to determine if there are less invasive biomarkers in serum and/or cerebrospinal fluid for the conditions’ understudied cerebrovascular phenotype.
Notably, 28% of the children followed in our cohort had a brain MRI performed for other reasons before any rheumatological diagnosis was made. Moreover, neuroimaging findings preceded the development of any dermatological abnormalities by at least 1 year in 20% of our patients. To our knowledge, no prior study has described a cohort of children with neuroimaging findings that significantly precede rheumatological diagnoses of PRS and/or ECDS. Further research in children with PRS and ECDS, including those with neuroimaging findings that precede rheumatological diagnoses, may help clarify the underlying processes that account for the temporal discrepancy appreciated in one-fifth of our patients. Independent of their pathophysiological mechanisms, these data raise the important point that neuroradiological findings may proceed any outward dermatological signs of rheumatological disease in children with PRS and ECDS. These clinical entities should therefore be considered part of the differential diagnoses by neurologists faced with unusual and unilateral neuroradiological findings.
While headaches and seizures are the most frequently reported neurological symptoms in patients with PRS and ECDS, 1,6,17 no prospective study has been initiated to determine the most efficacious way to treat either comorbidity in these patients. The current mainstays of rheumatological treatments for PRS and ECDS are anti-inflammatory and immunosuppressive agents, such as corticosteroids, methotrexate and mycophenolate mofetil. 18 -20 Use of immunosuppression specifically for headaches or seizures in children with PRS and/and ECDS is not currently clinically indicated given that many different therapies exist for both comorbidities with far less adverse effects. However, a prospective study of how immunosuppression affects headache burden, seizure control, and lesions on susceptibly weighted images before and after treatment initiation in patients otherwise requiring the treatment may provide clues into the pathophysiology of the neuroradiological findings and neurological complaints we describe here.
The current study is largely limited by its retrospective nature. Despite the study’s relatively large sample size, not all children with PRS and ECDS had neuroimaging completed. Recent consensus guidelines have suggested universal neuroimaging with MRI in scleroderma patients with facial involvement, 20 though prospective data are lacking. Moreover, given a lack of clear validated biomarkers, diagnosis of PRS and ECDS relied heavily on the clinical expertise of rheumatological subspecialists. Determining the presence or absence of neurological symptoms was also limited to whether these complaints were explicitly documented in clinical visit notes. Moreover, in no way can the neuroradiological findings discussed here be definitively attributed to PRS or ECDS given they can be plausibly found in asymptomatic individuals as well as patients with idiopathic headaches or seizures. Despite our institution’s large catchment area, the study was performed at a large, tertiary children’s hospital, and thus selection bias cannot be ruled out.
Conclusions
Neuroradiological findings were present in a subset of children with PRS and ECDS. T2-weighted hyperintensities and lesions on susceptibility-weighted images were the 2 most common abnormalities. Lesions were most often supratentorial, stable, unilateral, and ipsilateral to the patient’s dermatological findings. Many children in our cohort who endorsed a history of headaches and/or seizures had susceptibility lesions on their brain MRIs. Neurologists should consider PRS and ECDS in their differential diagnosis when confronted with unique, unilateral neuroradiological findings in a child, even if there are no clear dermatologic symptoms at initial presentation.
Footnotes
Author Contributions
RRS, KT, and KST designed the study. RRS, AF, and KT reviewed all neuroimaging. RRS and DG conducted the data analyses. RRS wrote the manuscript. RRS, DG, AF, KT, and KST revised the manuscript.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported financially by funding from the National Institutes of Health (AR073516 to KST) and the Nancy Taylor Foundation for Chronic Diseases (KST).
Ethical Approval
Ethical approval for this study was obtained from the University of Pittsburgh Institutional Review Board (Protocol #14080297).