Dr. H. Tiemeier is responsible for statistical analysis.
Address correspondence to Christian G. Bien, Epilepsy Center Bethel, Krankenhaus Mara, Maraweg 21, 33617 Bielefeld, Germany. E-mail: firstname.lastname@example.org
Purpose: Rasmussen encephalitis (RE) leads to progressive tissue and function loss of one brain hemisphere and often intractable epilepsy. This is the first randomized prospective treatment trial in RE.
Methods: Germany-wide, patients with suspected recent-onset RE were recruited and if eligible randomized to tacrolimus or intravenous immunoglobulins (IVIGs). A loss of motor function or hemispheric volume by ≥15% (in patients >12 years at disease onset: ≥8%) led to study exit. Untreated patients served as a historical control group.
Key Findings: Over 6.3 years, 21 patients with recent-onset RE were identified. Sixteen were randomized to tacrolimus (n = 9) or IVIG (n = 7). Immunotreated patients had a longer “survival” than the historical controls. Neither treatment was more efficacious than the other. Two tacrolimus patients experienced serious adverse events. No immunotreated but several untreated patients developed intractable epilepsy. No patient with refractory epilepsy became treatment-responsive under immunotherapy.
Significance: The countrywide incidence rate of diagnosed RE is estimated as 2.4 cases/107 people ≤age 18/year. Treatment with tacrolimus or IVIG may slow down tissue and function loss and prevent development of intractable epilepsy. However, immunotherapy may “arrest” patients in a dilemma state of pharmacoresistant epilepsy but too good function to be offered functional hemispherectomy. These compounds may therefore contribute to the therapeutic armamentarium for RE patients without difficult-to-treat epilepsies.
Rasmussen encephalitis (RE) is a rare inflammatory brain disease causing progressive cerebral hemiatrophy. It typically develops in children and is characterized by progressive hemiparesis, cognitive impairment and—usually— by refractory epilepsy (Rasmussen et al., 1958; Bien et al., 2005). Brain tissue studies revealed an oligoclonal granzyme B–mediated T-cell immunoreaction against neurons and astrocytes (Bien et al., 2002a; Bauer et al., 2007; Schwab et al., 2009).
Robust data on incidence and most effective treatment of RE are missing (Bien & Schramm, 2009). There is broad consensus (but no trial evidence) that hemispherectomy in one of its modern variants (HE; Schramm, 2002) is highly effective in eliminating the epileptic seizures (Bien & Schramm, 2009). HE is usually reserved for patients in the residual disease stage with dense hemiparesis and with language not localized in the affected hemisphere. Since the late 1990s, almost all patients diagnosed with RE prior to reaching such a residual stage are treated by some form of long-term immunotherapy (Granata et al., 2003). Most clinicians use intravenous immunoglobulins (IVIGs) or oral tacrolimus as long-term agents in RE. For both approaches, small-scale open case studies have provided promising results (Hart et al., 1994b; Bien et al., 2004). Long-term steroid treatment is often limited by intolerable side effects (Bahi-Buisson et al., 2007).
Herein, we present a prospective Germany-wide study assessing patients with incident RE. The patients were randomized to tacrolimus or IVIG and compared to historical controls who were not undergoing immunotherapy. Tacrolimus inhibits calcineurin and thereby the transcription of interleukin 2 in T cells. As a result, T cell proliferation and activation is impeded (Scott et al., 2003). The mechanism of action of IVIG is complex; it interferes with humoral autoimmunity, but may also inhibit T cell function, for example, by reduction of interleukin 2 and interferon γ (Kazatchkine & Kaveri, 2001).
Our study aimed at answering the following three questions:
1 What is the incidence of RE?
2 How is the disease course under continuous immunotherapy in comparison to untreated historical controls?
3 Can the trial demonstrate superior efficacy of IVIG or tacrolimus?
Patients and Methods
A Germany-wide initiative to refer patients with the suspected or confirmed diagnosis of RE to the Department of Epileptology, University of Bonn, for this study was launched in 2002. It addressed the tertiary centers for neurology, neuropediatrics, and epileptology in this country. A consensus on the study protocol was reached among participating colleagues. Colleagues were continuously invited to transfer RE patients by publications in German journals (Bien, 2003; Bien & Elger, 2005; Bien, 2008) and talks on national conferences.
Inclusion criteria (at least two had to be present):
1 History of epilepsia partialis continua or progressive hemiparesis.
2 Progressive cerebral hemiatrophy on serial magnetic resonance imaging (MRI) studies.
3 Biopsy evidence of T-cell dominated encephalitis with activated microglial cells (typically, but not necessarily forming nodules) and reactive astrogliosis; numerous macrophages, B cells, or plasma cells or positive signs of viral infections (obvious viral inclusion bodies or immunohistochemical demonstration of viral protein) excluded the diagnosis of RE.
“Progressive” means that at least two sequential clinical examinations or MRI studies document increasing deficits or tissue loss.
These criteria were incorporated as “part B” diagnostic criteria into the European consensus statement on RE (Bien et al., 2005).
1 Onset of acute disease stage >12 months ago.
2 Advanced cerebral hemiatrophy as indicated by a hemispheric ratio (HR; Bien et al., 2002b) <80% (<90% in patients ≥12 years); HR is the ratio of the pixels of the affected and the unaffected hemisphere on planar MR images. It is computed for defined coronal and axial slices, and is given as the mean of these two values.
3 Neuroradiologic signs of a bihemispheric encephalitis.
4 Immunologic pretreatment with >3 weeks of corticosteroids or tacrolimus, or >1.2 g/kg IVIG, or >5 plasma exchanges or immunoabsorptions within the last 3 months.
5 Patient already in residual stage, that is, stable neurologic deficit for >6 months.
6 Wave-like course with history of repeated remissions.
7 Infectious disease as a contraindication to an immunosuppressive therapy.
8 Known paraneoplastic encephalitis.
In the absence of an incidence figure and of quantitative immunotherapy efficacy data, a power calculation was impossible. We limited the inclusion period to 6 years, but planned an extension until at least 16 patients had been randomized. This was based on criteria of feasibility (the inclusion cannot be continued endlessly) and a power calculation that a very strong effect of the intervention (five times longer time to event) could be detected. We were aware that this was not a very likely scenario as it corresponds to complete cure by the intervention. With 72 patients, a more realistic effect of a two times shorter time to event would have been detected (details and assumptions of power calculation: significance level 0.05, two-sided p-value, accrual time 84 months, median failure time 18 months).
Standard protocol approvals, registrations, and patient consents
The Bonn University Ethics Committee approved the study, which was conducted according to the Declaration of Helsinki. The study was registered as NCT00545493 at http://www.ClinicalTrials.gov. Written informed consent was obtained from the parents of all patients who were participating in the study (consent for research).
Each participating patient was randomized to receive either tacrolimus capsules (Prograf; Astellas Pharma, Munich, German) or IVIG infusions (Octagam 5%; Octapharma, Langenfeld, Germany). Tacrolimus doses were chosen to achieve the following blood trough levels (measured by Dimension Clinical Chemistry Systems Tacrolimus Assay; Siemens Healthcare Diagnostics GmbH, Deerfield, IL, U.S.A.): 12–15 ng/ml during months 1–6; 5–10 ng/ml during months 7–12; and 5–8 ng/ml thereafter. In only three of nine tacrolimus-treated patients, all blood levels were within or above the intended ranges. This was mainly due to the interaction with anti-epilepsy drugs that induce liver enzyme activity (see Table S1). IVIG was given initially on three consecutive days at 0.4 g/kg body weight per day, thereafter once 0.4 g/kg every month, and after 12 months 0.4 g/kg every 2 months. Antiepileptic pharmacotherapy was at the discretion of the treating physicians. For phases of seizure exacerbation, additional steroid pulses were permitted. Patients were stratified into those <12 years at onset of the acute disease stage and those being ≥12 years—because older patients usually have a milder disease course (Hart et al., 1994a, 1997; Bien et al., 2002b). To conceal randomization, a computer-generated sequence of treatment allocations was sealed in numbered envelopes and locked in CGB’s office. The envelope with the lowest number in the appropriate strata bunch was opened when written informed consent had been given.
Follow-up and exit criteria
Patients were assessed clinically during the first year at two-month intervals, and thereafter every 4 months (±2 weeks). At each visit, Motricity index (MI) (Demeurisse et al., 1980; Wade & Hewer, 1987a,b; Collin & Wade, 1990) was determined by a clinician unaware of individual treatments. MI measures motor abilities of one (impaired) side of the body—usually in stroke, but also in other brain diseases, including those with inflammatory etiology (Soyuer et al., 2006; Freivogel et al., 2009; Gijbels et al., 2011). It relies on performance categories of the Medical Research Council scale applied to six movements (fingers – elbow – shoulder – hip – knee – ankle, details in Table S2). Scores <100 indicate motor function impairment. Brain MRI studies were done at 0, 2, 4, 6, and 12 months and then every 12 months (±2 weeks) to determine the HR. Values <100% indicate atrophy of the affected hemisphere. A formal evaluation of neuropsychology data is in preparation.
As historical comparators for the effectiveness of immunotherapy we used seven untreated historical control patients. They satisfied the following criteria to ensure comparability with the prospective study population: disease duration and HR at first assessment in the range required for inclusion into the prospective study participants; intervals between MRI studies <3 years (so that the time point of deterioration could be clarified with sufficient precision), and absence of the above-named exclusion criteria. Their course was retrospectively evaluated from chart review. HR was determined for study purposes afterwards. All available follow-up dates were used.
Fasting blood tests were done at baseline and at each clinical visit. In the tacrolimus group, also Epstein-Barr virus (EBV) and cytomegalovirus antibodies were determined. If titers were rising, virus load was determined. Tacrolimus patients underwent abdominal ultrasound studies to exclude lymphoma and nephrocalcinosis every 6 months during the first 12 study months—period of highest lymphoma incidence (Anonymous, 1998)—and thereafter every 12 months. At the same intervals, electrocardiographs were recorded for detection of cardiac abnormalities. Surveillance for potential unwanted effects outside the regular study visits were done by the local neurologists or neuropediatricians.
Loss of ≥15% of baseline HR or MI values (≥8% in the group ≥12 years) were taken as evidence of treatment failure. To exclude premature judgments on potentially reversible (e.g., postictal) motor deterioration, only an MI loss of ≥15%/8% documented twice with an interval of 1 month indicated treatment failure. Upon termination of the trial, full consensus on patients leaving the trial due to treatment failure (regarded as “exits”) versus other reasons (“censored”) was achieved between CGB and HT. The historical controls were studied for HR loss on all available serial MRIs.
Recently, characteristic treatment decision situations have been suggested as resulting from four possible combinations of the key aspects of RE, that is, severe epilepsy or no severe epilepsy, and severe or no severe neurologic deficit (Bien & Schramm, 2009). To translate this to the present trial, patients were retrospectively categorized by CGB, RS, and SK in consensus for baseline and most recent follow-up. A severe deficit (permitting offer of HE) was defined by loss of useful fine finger movements. Nonsevere deficit was defined as fine finger movements preserved. For refractory epilepsy, a recent formal definition was used (Kwan et al., 2010): failure of adequate trials of two tolerated, appropriately chosen and used anti-epilepsy drug (AED) schedules (whether as monotherapies or in combination) to achieve sustained seizure freedom. No refractory epilepsy was diagnosed if drug responsiveness was “undetermined” (because no sufficient treatment attempts had been undertaken yet) or if sustained seizure freedom had been achieved by antiepilepsy treatment (sustained seizure freedom defined as freedom from seizures for a minimum of three times the longest preintervention interseizure interval, determined from seizures occurring within the last 12 months, or 12 months, whichever is longer), or if a patient did not have epilepsy. This categorization resulted in groups defined by the combinations of “severe deficit + refractory epilepsy,”“nonsevere deficit + refractory epilepsy,”“severe deficit + no refractory epilepsy,” and “nonsevere deficit + no refractory epilepsy.”
For analysis of survival curves, Kaplan-Meier curves were established; we tested for significant differences using the log-rank (Mantel-Cox) test. Two-sided Fisher’s exact or Mann-Whitney U tests were used as appropriate. p-Values < 0.05 were considered significant.
Between October 1, 2002 and March 24, 2009 (prolongation of the initially planned inclusion period of <3 months), the intended minimum number of 16 patients was enrolled. The follow-up period was terminated according to the trial protocol 1 year later. Demographic, clinical, and neuroradiologic data of the patient groups and the historical control group are summarized in Tables 1, 2, and S3. Fourteen patients underwent open brain biopsy; in six of seven control patients, tissue diagnosis was available. All patients had evidence of chronic encephalitis.
Continuous data are given as medians (ranges). One patient who received the initial dose of IVIG did not appear for follow-up.
aEmpty because due to retrospective nature of assessment, only due to HR loss possible.
bEmpty because no prospective trial.
Follow-up duration (months)
Reasons for leaving the trial
HR loss (exit criterion)
MI loss (exit criterion)
End of follow-up
Pts. with refractory epilepsy
EPC ever (additional no. of pts. compared to inclusion)
Pts. with additional steroid pulses
Pts. proceeding to HE
Incidence and enrollment
Twenty-one patients with new-onset RE were identified over the inclusion period of 6.3 years, that is, 3.3 per year. Three of the 21 patients were ≥12 years at onset of the acute disease stage, the oldest one being 15.5 years. With 13.7 million inhabitants age 18 or younger in Germany (Statistisches Bundesamt, 2010), one can estimate an incidence rate of 2.4 cases per 107 inhabitants age 18 or younger per year. The distribution of patients’ hometowns is in congruence with the general population density in Germany (not shown). Details on the enrollment process are given in Fig. 1. Figure S1 shows the accrual of patients with recent-onset RE over time.
Course under tacrolimus and IVIG treatment
Seven patients exited due to treatment failure; nine patients left the trial without having met an exit criterion. Ten, six, six, and 2 patients remained in the trial by 12, 24, 36, and 48 months, respectively. Patients receiving tacrolimus or IVIG had a superior survival (i.e., absence of exit due to treatment failure) if compared to the historical controls, p = 0.038 (Fig. 2A). This was true even though, in the historical control group, only the HR exit criterion could be applied (no MI data were available) and the intervals in between the MRI studies were variable (with the possibility of overestimation of their “survival period,” see Table S4). Seizure frequencies over the study periods are depicted in Fig. S2.
All controls proceeded from nonsevere to severe deficits, and five underwent HE for intractable epilepsy. The HE rate was lower in the immunotreated patients (4/16—this includes one patient receiving only the initial course of IVIG and then not appearing for further treatment who finally underwent HE in another center). Neither this difference nor the survival difference between the tacrolimus and IVIG arms was significant (Fig. 2B).
The categorization of patients before treatment and at the end of the study reveals interesting results (Fig. 2C). None of the nine initially AED-resistant patients became responsive under continuous immunotherapy, but only one of seven patients without refractory epilepsy became treatment resistant (this was the one dropping out after only one course of IVIG). In contrast, three of five untreated historical control patients who were not pharmacoresistant at first investigation became refractory.
Looking at the severity of motor deficits, all historical untreated controls (and also the patient leaving the study after the first IVIG administration) proceeded from a nonsevere to a severe deficit; this occurred in only 8 of 13 immunotreated patients who had a nonsevere motor deficit at inclusion. Conversely, 5 of 13 remained in the state of a nonsevere deficit. Two of these patients with a mild deficit only still had no refractory epilepsy (i.e., a favorable situation). This leaves four immunotreated patients with AED-resistant epilepsy “arrested” with neurologic function worth of protection, thereby making the option of effective seizure treatment by HE virtually impossible. In none of these cases was there definite or probable language localization in the affected hemispheres, so it was only the preserved motor function that precluded HE.
Safety of tacrolimus and IVIG treatment
Serious adverse events (SAEs)
Two patients—both from the tacrolimus group—experienced SAEs that led to termination of this treatment. One had a febrile infection necessitating inpatient treatment with intravenous antibiotics and fluid substitution 12 months after onset of tacrolimus treatment (the patient recovered fully from this), one developed asymptomatic EBV viremia after 43 months into the study.
Six of nine tacrolimus patients had a total of 10 episodes of infections: coryza with cough, n = 2; gastroenteritis, n = 2; stomatitis, n = 1; upper airway infection, n = 1; pneumonia, n = 1; appendicitis, n = 1; febrile infection not otherwise specified (SAE), n = 1; EBV viremia (SAE), n = 1. In comparison, one of six IVIG patients (no significant difference to tacrolimus group) experienced two infectious episodes (purulent infection of big toe, n = 1; febrile infection not otherwise specified, n = 1).
Abnormal laboratory values
These were more common in patients on tacrolimus than in patients receiving IVIG (see Table S5).
CNS side effects
Three tacrolimus patients had nausea with vomiting, nausea with headache, or appetite loss related to blood levels of 22.3, 3.6, and 18.9 ng/ml. These side effects resolved on lowering the blood levels to 12.8, 3.0, and 5.8 ng/ml, respectively.
None of the tacrolimus-treated patients developed cardiac abnormalities, lymphoma, or nephrocalcinosis during the trial.
This countrywide trial in recent-onset RE gives the first available incidence rate estimate for this disorder: 2.4 cases per 107 people age 18 or younger per year. It cannot be excluded that the trial underestimates the real incidence; some cases, for example, less severe ones, may have escaped attention. Several arguments, however, suggest that we did not miss a relevant number of at least the patients presenting to medical institutions with higher levels of care: all efforts were made to encourage colleagues to transfer patients with suspicion of this difficult and rare diagnosis to the study center. There was a steady, almost linear, recruitment, and the spatial distribution of the patients’ hometowns reflects the population density of Germany. A comparison with the largest available RE series from Canada (47 cases age 18 or younger at onset from 1950 to 1987; one must note, that all were surgical cases, that the mean disease duration was 5.2 years, and that some patients may have been transferred from abroad) with a mean population of about 20 million in this period (Statistics Canada, 2012) and an approximated proportion of one fourth for those aged 18 years or younger gives a very similar incidence estimate of 2–3 cases per 107 people in this age range per year (Oguni et al., 1991).
Patients receiving immunotherapy had a delayed deterioration compared to untreated historical controls. Nevertheless, only 38% and 13% of the randomized patients were still at risk of treatment failure (i.e., in the trial) by 3 and 4 years after initiation of immunotherapy. These results may be marginally better than expected from small numbers of historical controls. Historical controls are certainly not ideal as time trends, and changes in patient care may confound the comparison. Compared with the randomized patients, the historical controls had shorter disease duration since onset of the acute stage and a longer duration of the prodromal stage, had higher HR values, were more often female, and were older. Despite these limitations and differences (nonsignificant, but potentially meaningful), a placebo-control was considered not feasible and even potentially unethical.
This study is underpowered to determine whether tacrolimus or IVIG may be more effective. Of interest, the Kaplan-Meier survival curves of the tacrolimus cohort suggests a delayed effect of this drug in terms of stabilization after 8–10 months, whereas IVIG seems to be effective earlier but to lose efficacy later on. Also noteworthy, the only two patients retained with the most favorable outcome—no refractory epilepsy and little neurologic deficit—were in the IVIG group. Serious and nonserious side effects and abnormal laboratory values were less frequent in the IVIG group but differences were not significant. Tacrolimus is more difficult to dose (as reflected by the limitation that not in all cases the intended blood levels could be achieved); on the other hand, IVIG requires regular IV administration, which may cause more disruption of normal life than intake of capsules. To improve efficacy, a combination of both compounds may be a worthwhile next trial step. Other potentially promising compounds may be the biologicals recently evaluated in multiple sclerosis (Barten et al., 2010).
An unresolved issue is the pharmacologic treatment of refractory epilepsy in RE. Immunologic treatment did not make intractable epilepsy responsive. On the one hand, patients without intractable epilepsy did not develop refractory epilepsy under immunotherapy. However, transgression to refractoriness was observed in three of five initially nonresistant historical control patients not undergoing immunotherapy. Therefore, the apparent inability of IVIG and tacrolimus to reverse pharmacoresistance leads to a Pyrrhic victory in patients having developed intractable epilepsy before onset of immunotherapy. Four out of nine become arrested in a state with useful motricity but a persistence of intractable and often disabling epilepsy.
These observations lead to the following suggestions for long-term management of recent-onset RE—in congruence with earlier suggestions (Vining et al., 1997): Families presenting a child with still useful fine finger movements but refractory epilepsy should be informed early about the poor prognosis regarding seizure-freedom except by HE. The deficits to be expected after HE should be weighed against the severity of the epilepsy in the given individual: Whereas the prediction of postoperative aphasia in a still communicable child with RE of the language dominant hemisphere will mostly preclude surgery, the perspective of hemianopia and a fixed dense hemiparesis with preserved walking abilities may be an acceptable price for seizure freedom (Bien & Schramm, 2009). An immunotherapy trial before a surgical decision is probably acceptable anyway. Patients with useful function but without refractory seizures, however, may be offered long-term tacrolimus or IVIG treatment.
Astellas Pharma GmbH, München, Germany (producer of Tacrolimus capsules), and Octapharma GmbH, Lingen, Germany (producer of Octagam) supported the study.
Dr. Bien served on a scientific advisory board of UCB, Germany, and Eisai, Germany; undertook industry-funded travel with support of Eisai, UCB, Desitin, and Grifols (all Germany); and obtained honoraria for speaking engagements from Eisai, GlaxoSmithKline, UCB, and Desitin (all Germany). Dr. Tiemeier obtained funding for scientific research projects from the VIDI scheme (NOW-ZonMw 017.106.370), and from the Sopohia Children’s Hospital Research Foundation (SSWO-2009-628). Dr. Kuczaty served on a scientific advisory board of UCB, Germany. Dr. Urbach served as an editor of the following journals: Clinical Neuroradiology and Neuroradiology. Dr. Bast served on scientific advisory boards of UCB, Eisai, and Desitin (all Germany), and as an Assistant Editor of Brain Top. Dr. Kurlemann served on a scientific advisory board of UCB, Germany, and received honoraria for speaking engagements from Desitin, UCB, and Eisai (all Germany). Dr. Rona received honoraria for speaking engagements from Medtronic, Inc. Dr. Schubert-Bast received honoraria for speaking engagements from Desitin, UCB, and Eisai (all Germany). Dr. Elger did industry-funded travel supported by Desitin, Pfizer, and UCB (all Germany); he served as an Editor of Epilepsy & Behavior; he received honoraria for speaking engagements from the Management Forum Pfizer; he obtained research support from the government entity Deutsche Forschungsgemeinschaft and a grant from the University of Bonn. The remaining authors have no conflicts of interest to disclose. 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.