Clinical application and evaluation of the Bien diagnostic criteria for Rasmussen encephalitis
Heather E. Olson,
Division of Epilepsy, Department of Neurology, Boston Children's Hospital, Boston, Massachusetts, U.S.A
Harvard Medical School, Boston, Massachusetts, U.S.A
Address correspondence to Heather Olson, Division of Epilepsy and Neurophysiology, Department of Neurology, Boston Children's Hospital, 300 Longwood Ave., Boston, MA 02115, U.S.A. E-mail: email@example.com
The 2005 diagnostic criteria for Rasmussen encephalitis (RE) are based on seizures, clinical deficits, electroencephalography (EEG), neuroimaging, and pathology (Brain, 128, 2005, 451). We applied these criteria to patients evaluated for RE and epilepsy surgery controls to determine the sensitivity, specificity, and positive and negative predictive values (PPVs, NPVs) using pathology as the gold standard.
We identified patients evaluated for RE based on medical records from 1993 to 2011. Fifty-two control patients with refractory epilepsy, unilateral magnetic resonance imaging (MRI) changes, and biopsies were selected from an epilepsy surgery database from matching years. Patients meeting all three of group A and/or two of three group B criteria were classified as meeting full criteria (positive). Patients not meeting full criteria were classified as negative. When available, pathology findings were re-reviewed with neuropathologists, and MRI imaging was re-reviewed with a neuroradiologist.
RE was considered in the differential diagnosis for 82 patients, of whom 35 had biopsies. Twenty patients met full criteria (positive) without another explanation, including seven for whom biopsy was required to meet criteria and one in whom another etiology was identified. Two patients met full criteria but had another explanation. Thirty-five met partial criteria (negative), of whom 14 had another etiology identified. Twenty-five met no criteria (negative). The diagnostic criteria had a sensitivity of 81% with four false negatives (criteria-negative, biopsy-positive) when compared to pathology as a gold standard. Five false positives (criteria positive, biopsy negative) had identifiable alternate diagnoses.
The 2005 Bien clinical diagnostic criteria for RE have reasonably high sensitivity and specificity and good clinical-pathologic correlation in most cases. We suggest modification of the criteria to allow inclusion of cases with well-described but less common features. Specifically we suggest making the diagnosis in the absence of epilepsia partialis continua (EPC) or clear progression of focal cortical deficits or MRI findings if biopsy is positive and two of the A criteria are met (B3 plus two of three A criteria). This would improve the sensitivity of the criteria.
Rasmussen encephalitis (RE) is a chronic T-cell–mediated disorder affecting one cerebral hemisphere that typically presents in childhood. Signs and symptoms of RE include cerebral hemiatrophy, focal cortical deficits, and refractory epilepsy (Bien et al., 2002a, 2005; Bien & Schramm, 2009). Formal diagnostic criteria were established by a panel of experts in 2005 (Bien et al., 2005). RE is described to have the following three stages: (1) prodromal period with mild signs and symptoms; (2) acute phase with progressive cortical dysfunction and frequent focal onset seizures, at times presenting with epilepsia partialis continua (EPC); and (3) residual stage with stable neurologic deficits and continuation of seizures. Atypical features of some reported cases included absence or delayed-onset seizures, dual pathology (i.e., cortical dysplasia or low grade tumor), and bilateral cerebral involvement (Chiapparini et al., 2003; Bien et al., 2005, 2007).
Pathologic findings are considered the gold standard for diagnosis of RE and are part of the clinical criteria (Bien et al., 2005). However, pathologic evaluation of patients is complicated by the patchy localization of the pathology and change in the pathologic manifestations over time (Bien et al., 2002b; Pardo et al., 2004).
To date, there has not been a study to assess the application of the Bien criteria in clinical practice. Our goals were the following: (1) to determine the correlation between 2005 Bien diagnostic criteria for RE and biopsy-proven cases, and (2) to evaluate the sensitivity and specificity of the Bien diagnostic criteria for RE.
Standard protocol approvals, registrations, and patient consents
This study was performed with the approval and in accordance with the guidelines of the institutional review board of Boston Children's Hospital (BCH).
Patient and control identification and data collection
To identify patients with suspected RE, the BCH medical records database was searched using the term “Rasmussen,” “Rasmussens,” or “Rasmussen's” between 1993 and 2011. We abstracted data from charts using a predetermined process, entered data into a database specifically designed for this study, and classified whether patients met 2005 Bien diagnostic criteria as outlined in Table 1 (Bien et al., 2005). Data extraction included demographics, side of pathology, and clinical presentation including cortical deficits and seizure semiology; progression of cortical deficits and seizures over time; progression of physical examination changes; development of EPC, magnetic resonance imaging (MRI), and electroencephalography (EEG) data, and biopsy results. Duration of patient follow-up at BCH ranged from a single clinic visit for a second opinion to >5 years.
Table 1. Clinical criteria for Rasmussen encephalitis
Patients need to meet either A or B criteria.
Progressive means that at least two sequential clinical examinations or MRI studies are required to meet the respective criteria.
T-cell–dominated encephalitis with activated microglial cells (typically, but not necessarily forming nodules and reactive astrogliosis
Numerous parenchymal macrophages, B cells, or plasma cells or viral inclusion bodies exclude the diagnosis of RE
MRI images from possible or definite RE cases, when available, were reviewed by a board certified pediatric neuroradiologist (SPP). Complete image sets were available for re-review in 25 (61%) of 41 cases meeting some or all criteria for RE without an alternate explanation. Images were evaluated specifically for unilateral or bilateral cortical/subcortical T2 signal change, unilateral or bilateral cerebral atrophy, signal change or atrophy of the caudate head, and ipsilateral or contralateral cerebellar atrophy. In addition, images were assessed for other central nervous system (CNS) pathology such as infarction, tumor, or malformations of cortical development.
Pathology slides from possible or definite RE cases, when available, were systematically reviewed by neuropathologists (PC and ML). Slides from biopsies or focal surgical resections were available for re-review in 17 (74%) of 23 cases meeting some or all criteria for RE without an alternate explanation and with a biopsy or pathology from surgery. Nineteen of the 42 cases in these categories did not undergo biopsy or surgery. No autopsy samples were available for re-review. The material consisted of routinely processed hematoxylin and eosin (H&E) sections, as well as select sections stained with Luxol Fast Blue, Kluver-Barrera staining, and Bielschowsky staining.
When cell markers were available (four cases), immunohistochemical staining was performed using anti-glial fibrillary acidic protein (GFAP) antibody (Ventana Medical Systems, Tucson, AZ, U.S.A.), anti-CD68 (Ventana Medical Systems), anti-CD3 (Ventana Medical Systems), anti-CD20 (Ventana Medical Systems), anti-CD4 (Ventana Medical Systems), anti-CD8 (Ventana Medical Systems), neuronal nuclear antigen (1:100 dilution; Chemicon, Temecula, CA, U.S.A.), and with anti-phosphorylated neurofilaments (1:2,000 dilution, SMI-31R; Covance, Emeryville, CA, U.S.A.) monoclonal antibodies. In brief, 5-mm tissue sections of formalin-fixed, paraffin-embedded tissue were used for immunohistochemical staining. The reactions were carried out in an automated immunohistochemistry instrument (Ventana Medical Systems). Antigen–antibody reactions were revealed with standardized development times. Cell markers were not available for cases with pathology before 2001, and inconsistently used after that time.
Features reviewed included lymphocytic infiltration in the parenchyma, perivascular spaces, and meninges differentiating T and B cells when possible microglial activation, astrocytosis, presence or absence of plasma cells or macrophages; and neuronal loss. Each was qualitatively rated as mild, moderate, or severe. We also evaluated samples for other pathology such as evidence of dysplasia or tumors. The pathologic findings were classified according to the Bien clinical criteria as well as according to previously published staging of cortical pathology in RE based on the review with study pathologists or, when slides were unavailable, on pathology reports or clinical notes (Pardo et al., 2004; Bien et al., 2005). When cell markers were unavailable we evaluated the criteria of T-cell–predominant encephalitis based on lymphocytic infiltrate without differentiation of B and T cells. We excluded inflammation related to recent surgery or subdural grids.
Controls were selected from an epilepsy surgery database, which includes all patients who underwent epilepsy surgery at BCH from 1993 to 2011. Included patients had refractory epilepsy, unilateral MRI abnormalities, and a biopsy reported in the BCH system. Controls were matched to cases based on time frame of biopsy, within 2 years during the same time frame. The age range of cases and controls were both 2–29 years. Patients with prior resective surgeries were excluded. Data were abstracted from control charts in the same manner as used for the case group above.
Cases and controls were classified according to the Bien 2005 diagnostic criteria and whether or not there was an alternative diagnosis. Longitudinal clinical, radiologic, electrophysiologic, and pathology data were reviewed as described earlier and incorporated to best classify patients based on all information available.
Statistical analysis included calculation of sensitivity, specificity, and positive and negative predictive values (PPV and NPV) of the Bien diagnostic criteria in patients compared to biopsy as the gold standard (consistent with the pathologic definition defined in Bien et al., 2005 criteria B3, Table 1).
A retrospective series of patients for whom Rasmussen encephalitis was considered in the differential diagnosis
One hundred eighty-seven patients were identified by the text search. A flowchart of case identification and available data is provided in Figure 1. Based on chart review, RE was considered in the differential diagnosis for 82 patients, of whom 35 had pathology specimens.
Considering the total 82 patients for whom RE was considered in the differential diagnosis, 20 patients satisfied the Bien clinical criteria for RE (clinical criteria positive) without another etiology identified, two satisfied the Bien criteria except that an alternate etiology was identified (ischemic stroke), 35 met some of the clinical criteria but did not satisfy sufficient part A or part B criteria to warrant a diagnosis of RE (clinical criteria negative), and 25 did not meet any of the criteria (typically mentioned in consideration of a new diagnosis of focal epilepsy). Outcome based on application of the clinical criteria for RE is depicted in Figure 2A.
Pathology samples were available in 18 of the 20 patients meeting clinical criteria for RE, of whom 7 patients would not have met the criteria without consistent histopathology (B3 in Table 2). Two met the criteria part A, 10 met the criteria part B, and eight met both the criteria part A and part B. Of the two without pathology available, one met the criteria A1–A3 and one met criteria B1 and B2 with progressive cortical deficits and progressive unilateral cortical atrophy. Seventeen of these patients were clinically diagnosed with RE by the clinicians who cared for them. For one patient, RE was a strong consideration. In two patients the clinical course over time deviated from that expected for RE, with a pattern of gradual improvement and lack of focal cortical deficits but no etiology other than RE identified.
Table 2. Demographic/clinical characteristics of cases and controls
Controls (N = 52)
Cases with RE considered, plus had a biopsy (N = 35)
Cases meeting Bien diagnostic criteria for RE, no alternate etiology (N = 20, 18 with biopsy)
Bilat, bilateral; F, female; L, left; M, male; N/A, not applicable; R, right; RE, Rasmussen encephalitis.
For controls versus RE cases, there was no significant difference between groups for age at biopsy, Mann-Whitney U-test for age p = 0.667. There was no statistically significant difference in gender ratio between groups, Fisher's exact test, p = 1.00. Side of pathology differed between groups with more left-sided pathology in controls and more right-sided pathology in RE cases, Fisher's exact test p = 0.039.
Gender, M:F ratio
1.3:1 (20 M, 15 F)
Age at biopsy (if done, and range)
Mean 9.2 years, Median 7.6 years, range 2.25–25.7 years
Mean 8.1 years, Median 7.4 years, range 1.08–28.8 years
Mean 10.1 years, Median 9.1 years, range 2.08–28.8 years
Side of pathology, L:R ratio
0.79:1, one bilateral (15L, 19R, 1 bilat)
1:2 (7L, 13R)
Of the 35 patients meeting some but not sufficient criteria, 14 had another etiology identified including malformations of cortical development (7), hemimegalencephaly (1), infectious meningoencephalitis (1 pneumococcal, 2 presumed acute infectious), and ischemic vascular events (3).
Considering only the 35 of 82 patients who had pathology available, 18 met criteria for RE without an alternate explanation, 12 had a clear alternate explanation (included above), and the remaining 5 did not have a clear diagnosis (Fig. 1).
A flowchart of results applying the clinical criteria for RE to the control group is provided in Figure 2B. Three of 52 patients in the control group satisfied the Bien diagnostic criteria for RE based on part A. They were all false positives, with inconsistent pathology, and had alternate explanations. The epilepsy etiology in control cases was as follows: malformation of cortical development (21), both malformation of cortical development and tumor (5), tumor (7), stroke (2), mesial temporal sclerosis (9), Sturge-Weber syndrome or isolated leptomeningeal angiomatosis (2), trauma (1), infectious meningoencephalitis (3), and unknown (2). Table 2 compares age at biopsy, male-to-female ratio, and side of pathology for the control subjects compared to the cases.
Sensitivity and specificity of the Bien diagnostic criteria for RE
Sensitivity and specificity were calculated based on all patients who had biopsies and who were evaluated using the Bien diagnostic criteria for RE (Table 1). This included 35 patients in the group of patients for whom RE was considered in the differential diagnosis and 52 control patients. The diagnostic criteria had a sensitivity of 81% with four false negatives (criteria-negative, biopsy-positive) and specificity of 92% with five false positives (criteria–positive, biopsy–negative). The PPV of the diagnostic criteria was 77% and the NPV was 94%. Calculations are described in Table 3.
Table 3. Statistical calculations
Bien diagnostic criteria +
Bien diagnostic criteria −
Calculation of sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of the Bien diagnostic criteria compared to the gold standard of biopsy. The total number of subjects with biopsy and assessed for Bien diagnostic criteria was 87, including 35 from the group assessed for RE and 52 from the control group of epilepsy surgery patients. Sensitivity = A/A + B = 81%. Specificity = D/C + D = 92%. PPV = A/A + C = 77%. NPV = D/D + B = 94%.
A = 17 (True positives)
B = 4 (False negatives)
C = 5 (False positives)
D = 61 (True negatives)
Four of the five false positives had another etiology identified, and the fifth had possible dual pathology with cortical dysplasia identified on biopsy but not on imaging. Other etiologies included hemispheric atrophy from old trauma (1), mesial temporal sclerosis with ipsilateral temporal lobe atrophy (1), Sturge-Weber syndrome (1), and perinatal left middle cerebral artery distribution infarction (1).
There were a total of four patients who did not satisfy the Bien clinical criteria for RE but had consistent biopsies, considered as false negatives (Table 4, Data S1). Three of the cases were diagnosed with RE by the treating clinicians, and for one the diagnosis remained under consideration.
Table 4. Summary of false-negative cases
Bien criteria met
Missing or atypical features
A1, A2, B3
Normal MRI; lack of EPC or progressive cortical deficits
A1, A3, B3
Bilateral EEG slowing and epileptiform activity, and bilateral seizure onsets; lack of progressive atrophy, progressive cortical deficits, or EPC
A1, A2, B3
Mild atrophy on MRI but not meeting A3 criteria and without progression; lack of progressive cortical deficit or EPC
Lack of cortical atrophy, lack of focal cortical deficit, lack of unilateral slowing on EEG, lack of EPC
Atypical features in RE cases that met criteria and had consistent biopsies
Nine cases (>50%) had atypical clinical, radiologic, and/or electrophysiologic features but still met the Bien clinical criteria and had consistent biopsies showing a T-cell–predominant meningoencephalitis. Atypical features are summarized in Table 5 and in the Data S1 of case histories. Figure 3 shows examples of typical and atypical MRI images. Pathology from one patient with dual pathology (cortical dysplasia and RE) is shown in Figure 4, and the case history is in the Data S1 of case histories.
Table 5. Summary of atypical features in cases of RE satisfying the Bien clinical criteria and with consistent biopsies
Atypical features or Bien clinical criteria positive, biopsy positive RE cases
Signal change without atrophy (2)
Ipsilateral (2) or contralateral (2) cerebellar atrophy
Lack of epilepsy (1)
Generalized seizures (3)
Bilateral slowing and epileptiform activity (5)
Generalized epileptiform activity and generalized seizures (i.e., spasms) (3)
Dual pathology with cortical dysplasia and RE (2)
Slowly progressive course (1)
We evaluated the clinical criteria developed in 2005 for diagnosis of RE (Bien et al., 2005). The Bien criteria constitute an important step in the initial assessment of RE, but further refinement to improve diagnostic accuracy in clinical practice might be needed. Diagnostic criteria often evolve over time as new information is gathered. Overall, the data show reasonably high sensitivity, specificity, PPV, and NPV. Although specificity is not a concern because false positives (criteria positive, biopsy negative) had an identifiable alternate diagnosis, an even higher sensitivity should be of utmost importance to avoid underdiagnosis of cases that may benefit from treatment. We therefore suggest tentative revisions that may be helpful to improve sensitivity. Specifically we recommend the following: (1) considering patients meeting criteria B3 plus two of three A criteria as positive and (2) expanding the pathology criteria to allow for stages of RE.
Notably, three of the four false negative cases (criteria negative, biopsy positive) were diagnosed with RE by treating clinicians. One patient underwent hemispherectomy, one was receiving intravenous immunoglobulin (IVIG) treatments, and for one patient, physicians were considering surgery versus immunomodulatory therapy at last follow-up. This suggests that strict application of the criteria may miss some patients, some of whom may benefit from RE-directed treatments. Based on these cases, it may be helpful to make the diagnosis even in the absence of EPC or clear progression of focal cortical deficits or MRI findings if biopsy is positive and two of the A criteria are met (B3 plus two of three A criteria).
The atypical features seen in our series of RE cases are similar to those reported previously, including dual pathology most frequently with cortical dysplasia in approximately 10% of RE patients, bilateral slowing and epileptiform activity on EEG, absent or delayed onset of epilepsy, and movement disorders including hemichoreoathetosis related to basal ganglia involvement (Frucht, 2002; Bien et al., 2005, 2007; Rhodes et al., 2007; Ferrari et al., 2011). A slowly progressive course is more typical of adult-onset RE but was seen in one childhood onset case in our series (Bien et al., 2002a, 2005). Cerebellar atrophy was ipsilateral in some cases and contralateral in others, as reported previously (Chiapparini et al., 2003; Cianfoni et al., 2010). Generalized seizures were seen in three cases. Although not well documented in RE, focal pathology can at times present with a generalized epilepsy, particularly in the form of epileptic spasms as seen in one patient in our series and one prior case report (Ferrari et al., 2011). Generalized EEG seizures may resolve or improve after focal epilepsy surgery or hemispherectomy in patients with selected focal lesions (Yum et al., 2011; Moseley et al., 2012). Many of these features may potentially prevent patients from meeting the current clinical criteria for RE, as shown in our series. Revision of the criteria as above would allow for inclusion of patients with these less common, but now well described, clinical features including bilateral or generalized seizures, bilateral slowing on EEG, or atypical imaging as seen in some RE cases.
Because diagnostic blood or cerebrospinal fluid (CSF) biomarkers for RE have not been identified, pathology is an important aspect of the diagnosis of RE and is included within the clinical diagnostic criteria. The inclusion of biopsy in the criteria may minimize false positives or false negatives because having a consistent or inconsistent biopsy increases or decreases, respectively, the chances of meeting the clinical criteria. Only one third of RE patients required biopsy for diagnosis, however. This suggests that biopsy may not be required for diagnosis in all patients if the clinical history is strongly suggestive, whereas in the past biopsy has been considered the gold standard. Based on positive predictive value of 77%, caution should be taken, however, when diagnosing a patient based on clinical criteria without biopsy confirmation.
Pathologic involvement in RE is a patchy process with fluctuating presentation in localization and time, as well demonstrated in a study of the pathology of a series of 45 RE patients who underwent hemispherectomies (Bien et al., 2002b; Pardo et al., 2004). Therefore, especially when considering small and at times nonlesional biopsies, the presence or absence and the degree of inflammation may be highly variable. Four pathologic stages of RE have been defined, of which T-cell infiltration is minimal to rare in stages three to four of the disease process and can be mild at the onset. If patients undergo biopsy outside of the stage one to two window, it is possible that they could have RE but not meet the B3 pathology criterion, which is heavily based on the T-cell–predominant encephalitis. Future revisions of the criteria may consider including a broader spectrum of pathology based on staging.
Our findings need to be interpreted in the setting of data acquisition and inherent biases, including study design, referral, and selection bias because the study was done in a tertiary care referral center as well as information bias. A simpler study design would have been to review only biopsy-confirmed cases of RE and determine if they met clinical criteria. We chose this study design, evaluating cases for whom RE was considered in the differential diagnosis compared to a control group of epilepsy surgery patients, because biopsy information is frequently not available early in the clinical course and we felt that evaluating the data longitudinally from both the clinical perspective and the pathologic perspective added valuable clinical insights. The screening method of searching charts by keyword introduces selection bias, and may have missed cases for which RE was considered in the differential diagnosis but was not specifically mentioned in the medical record. We added a control group to adjust for some of these effects to the best of our abilities. Small sample size limits statistical power as is typical of rare diseases. Future multicenter studies with larger sample sizes would be beneficial.
Our chart review covered a broad longitudinal time frame. As a result, imaging techniques and quality were highly variable and several MRI studies from before 2000 were not available for re-review with radiology. Pathology techniques also varied with cell markers for B and T cells only available in select recent cases. Charts were not only reviewed cross-sectionally but also longitudinally, and this may have provided potentially additional information that was not available during a single clinic visit. It is therefore possible that we overestimated sensitivity and specificity of clinical criteria, in particular in patients presenting early in the course of the disease. RE is certainly a disorder for which diagnostic accuracy increases with information on evolution of clinical features over time, and the current criteria do incorporate temporal evolution. There are no previous attempts to validate the diagnostic criteria for RE and this is the best available data on diagnostic sensitivity and specificity.
We thank Dr. Matt Gregas, statistician, for statistics support during this project.
Dr. Olson, Dr. Lechpammer, Dr. Prabhu, Dr. Ciarlini, Dr. Poduri, Dr. Gooty, Dr. Anjum, and Dr. Gorman report no disclosures.
Dr. Loddenkemper serves on the Laboratory Accreditation Board for Long Term (Epilepsy and Intensive Care Unit) Monitoring, on the Council of the American Clinical Neurophysiology Society, on the American Board of Clinical Neurophysiology, as an Associate Editor for Seizure, and performs video-electroencephalography long-term monitoring, electroencephalography, and other electrophysiologic studies at Boston Children's Hospital and bills for these procedures. He receives support from the National Institutes of Health/National Institutes of Neurological Disorders and Stroke, a Career Development Fellowship Award from Harvard Medical School and Boston Children's Hospital, the Payer Provider Quality Initiative, the translational research project at Boston Children's Hospital, The Epilepsy Foundation of America (EF-213583 and EF-213882), the Center for Integration of Medicine and Innovative Technology, the Epilepsy Therapy Project, the Pediatric Epilepsy Research Foundation, CURE (Citizens United for Research in Epilepsy), the American Epilepsy Society, and investigator-initiated research grants from Lundbeck and Eisai. He also received travel support from the RE Children's Project. We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.