Article Text

Download PDFPDF

Limbic encephalitis in children and adolescents
  1. E Haberlandt1,
  2. T Bast2,
  3. A Ebner3,
  4. H Holthausen4,
  5. G Kluger4,
  6. R Kravljanac5,
  7. J Kröll-Seger6,
  8. G Kurlemann7,
  9. C Makowski8,
  10. K Rostasy1,
  11. E Tuschen-Hofstätter9,
  12. G Weber10,
  13. A Vincent11,
  14. CG Bien12
  1. 1Medical University of Innsbruck, Clinical Department of Pediatrics IV, Division of Neuropediatrics and Inherited Metabolic Diseases, Innsbruck, Austria
  2. 2University Children's Hospital, Heidelberg, Germany
  3. 3Epilepsy Centre Bethel, Clinic Mara, Bielefeld, Germany
  4. 4Neuropediatric Department, BHZ Vogtareuth, Vogtareuth, Germany
  5. 5University of Belgrade, Institute for Mother and Child Health, Belgrade, Serbia
  6. 6Swiss Epilepsy Centre, Zurich, Switzerland
  7. 7University of Muenster, Department of Neuropediatrics, Muenster, Germany
  8. 8Munich University of Technology, Department of Pediatrics, Munich, Germany
  9. 9St. Marien-Hospital Bonn, Pediatric Department, Bonn, Germany
  10. 10Klinikum Duisburg, Department of Pediatrics, Duisburg, Germany
  11. 11University of Oxford, L6 West Wing and Weatherall Institute of Molecular Medicine, John Radcliffe Hospital, Oxford, UK
  12. 12University of Bonn Medical Centre, Department of Epileptology, Bonn, Germany
  1. Correspondence to Dr Christian G Bien, Department of Epileptology, University of Bonn Medical Centre, Sigmund-Freud-Str. 25, 53105 Bonn, Germany; christian.bien{at}


Objective Limbic encephalitis is rare in people <18 years of age and rarely given a formal diagnosis.

Design Retrospective study on presentation and outcome of children and adolescents with the clinico-radiological syndrome of limbic encephalitis tested for specific neuronal autoantibodies (Abs) over 3.5 years.

Setting Assessment, diagnosis, treatment and follow-up at 12 neuropaediatric and neurological departments in Europe, with Abs determined in Bonn, Germany and Oxford, UK.

Patients Ten patients <18 years of age who presented with a disorder mainly affecting the limbic areas of <5 years' duration with MRI evidence of mediotemporal encephalitis (hyperintense T2/FLAIR signal, resolving over time).

Results Median age at disease onset was 14 years (range 3–17). Eight patients had defined Abs: one each with Hu or Ma1/2 Abs, four with high titre glutamic acid decarboxylase (GAD) Abs, two of whom had low voltage-gated potassium channel (VGKC) Abs and two with only low titre VGKC Abs. A tumour was only found in the patient with Hu Abs (a neuroblastoma). After a median follow-up of 15 months with corticosteroid or intravenous immunoglobulin treatment, starting after a median of 4 months, two patients recovered, eight remained impaired and one died.

Conclusions Limbic encephalitis is a disease that can occur in childhood or adolescence with many of the hallmarks of the adult disorder, suggesting that both result from similar pathogenic processes. Since most of the cases were non-paraneoplastic, as now also recognised in adults, more systematic and aggressive immunotherapies should be evaluated in order to improve outcomes.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Limbic encephalitis is a clinico-pathological entity with mediotemporal lobe symptoms caused by inflammatory limbic lesions. Symptoms consist of impairment of recent memory, temporal lobe seizures and affective disturbances. Limbic encephalitis can occur as a paraneoplastic1 or non-paraneoplastic condition2 3 and can be subclassified by the presence of specific neuronal autoantibodies (Abs). These include onconeural Abs which are associated with paraneoplastic disease,4 and Abs to the voltage-gated potassium channel complex (VGKC)3 or glutamic acid decarboxylase (GAD) which are usually non-paraneoplastic in origin.5

In most cases, limbic encephalitis is a disorder of adult life. The largest available series on paraneoplastic limbic encephalitis (PLE) gives an age range down to 11 years.1 Case reports on only five paediatric patients, ranging in age from 5 to 16 years, have been published in the English literature and four of these had PLE.6,,10 Abs have only rarely been reported in children,8 10 and little systematic data on patients' presentation and outcomes are available. A series of 10 European patients is presented here.

What is already known on this topic

  • Limbic encephalitis is an inflammatory brain disorder of paraneoplastic or non-paraneoplastic origin, well recognised in adults, causing memory deficits, temporal lobe seizures or affective disturbances.

  • Limbic encephalitis is rarely diagnosed formally in patients <18 years of age.

What this study adds

  • Nine of the 10 patients in this series had limbic encephalitis without evident tumours; eight were positive for autoantibodies known to be associated with adult-onset limbic encephalitis.

  • Outcome was unfavourable in 8/10 patients; in adults, steroid or intravenous immunoglobulins may be effective.

Patients and methods

All patients <18 years of age at disease onset who were brought to CGB's attention for Ab testing and discussion of diagnosis between March 2006 and August 2009 and meeting the following diagnostic criteria of limbic encephalitis were included: (1) signs and symptoms indicating a disorder mainly affecting the limbic areas (impaired recent memory, temporal lobe seizures or affective abnormalities), present for ≤5 years; and (2) MRI evidence of mediotemporal encephalitis (ie, hyperintense T2/FLAIR signal, resolving over time). Other potential diagnoses were fully explored and excluded. Patients with N-methyl-d-aspartate-receptor (NMDAR) Abs and the typical cortico-subcortical encephalopathy referred to as anti-NMDAR encephalitis11 without predominant limbic dysfunction were not included here (during this time period, two such patients in this age group were identified).

All data were provided by the treating institutions. MRIs from presentation and last available follow-up were re-evaluated. Right and left mediotemporal lobe structures (hippocampi, amygdalae) were classified as atrophic or non-atrophic. Corticosteroid doses were expressed as methylprednisolone-equivalents (prednisone doses multiplied by 0.8, dexamethasone doses by 512).

Ab testing was performed in Bonn and Oxford. Sera and cerebrospinal fluids (CSFs) were first tested by immunohistochemistry (IHC) in Bonn for binding to neuronal antigens on paraformaledhyde-perfused cryoprotected rat brain to detect high-titre GAD Abs, onconeural Abs4 and Abs to hippocampal neuropil, which includes VGKC, NMDAR and other undefined antigens. All sera and CSF samples showing typical staining for GAD Abs were titrated to determine the lowest dilutions giving a staining pattern and confirmed by radioimmunoprecipitation assay (RIA) using 125I-GAD (RSR, Cardiff, UK; normal values ≤1 U/ml). In addition, all 10 serum samples were tested for VGKC Abs by RIA (normal values in adults <100 pmol/l; 100–150 pmol/l designated ‘low positive’, 150–400 pmol/l designated ‘positive’; >400 pmol/l designated ‘high positive’). All sera were also tested for onconeural Abs to Hu, Ma1, Ma2, amphiphysin and CV2/CRMP5 by immune-dot-blot (Ravo Diagnostika, Freiburg, Germany). Test details are given in Malter et al.5

One patient with GAD Abs and low positive VGKC Abs was included in a case report previously, before the GAD Ab positivity was known.13 Two other patients with GAD Abs have been included in a series on principally adult limbic encephalitis with GAD Abs (one in the core study group, the other in an additional group of four patients who underwent operations).5

No diagnostic or therapeutic interventions were designed or performed for the purposes of this study. Since this is a retrospective anonymous description of clinically studied and treated patients, no ethics committee approval was required.


Ten patients from 12 institutions in four European countries were identified as having limbic encephalitis according to their clinical presentation and radiological evidence of mediotemporal lobe inflammation (see Patients and methods section). One institution contributed two patients, and two patients were treated first by a neuropaediatric and later by a neurological department. Clearly, patients with this condition may not have been included because the treating physicians were unaware of this diagnostic possibility in childhood, assumed a viral aetiology and did not consider sending the serum for analysis, or because of lack of adequate neuroimaging. Demographic data, neurological, neuroradiological and laboratory findings are summarised in table 1. A characteristic MRI course is given shown in figure 1.

Figure 1

Serial brain MRIs (FLAIR) of a girl with manifestation of limbic encephalitis at the age of 15 years (patient 6). The patient was highly positive for GAD antibodies and low positive for VGKC antibodies. She underwent a right anteromedial temporal lobe resection for intractable seizures 6.3 years after disease onset. Histopathology revealed chronic encephalitis and segmental loss of hippocampal pyramidal cells with gliosis (ie, hippocampal sclerosis).

Table 1

Demographic data, presentation, treatment and follow-up

Three patients had potentially relevant medical histories: in one girl (Ab negative limbic encephalitis at age 13.5), a rolandic epilepsy had been diagnosed and successfully treated with sulthiame at the age of 6; in a 3.5-year-old girl low positive for VGKC Abs, mild psychomotor retardation had been noted 1 year earlier; and in a 7.5-year-old girl with GAD Abs, type I diabetes was diagnosed 2 years before onset of the limbic encephalitis.

Two patients qualified as having paraneoplastic disease because of the combination of the classical syndrome of limbic encephalitis and the presence of onconeural Abs.4 In one of them, a 3.5-year-old girl with cerebellar degeneration and brainstem encephalitis in addition to limbic encephalitis, a tumour was identified. Even though an abdominal ultrasound had been unremarkable, the detection of Hu Abs prompted an abdominal MRI which demonstrated a mass that was shown after resection to be a ganglioneuroblastoma. In the other patient with PLE, a 13.5-year-old girl with definite evidence of Ma1/2 Abs (both on IHC and by blot), no tumour was found despite several extensive searches including CT of thorax and abdomen, bronchoscopy, gynaecological investigations and bone-marrow puncture. She continues to be observed.

Of the remaining eight patients, six had Abs detected. Four had high-titre GAD Abs (IHC titres 1:8000–1:128 000; >5000 U/ml on RIA; CSF range: 1:250–1:8000). Four patients had serum VGKC Abs in the low positive or positive range (143–310 pmol/l), two of them in conjunction with high-titre GAD Abs. Since VGKC assays are seldom requested in children, we also tested 11 sera from neuropaediatric patients with non-immunological disorders (median age 7.5 years, range 1–15 years); VGKC Ab titres ranged from −28 to 50 pmol/l (mean +3 SD 88 pmol/l), similar to adult controls (<100 pmol/l)). Of the eight patients without onconeural Abs, seven underwent tumour searches (all but one thoraco-abdominal CT; six of seven girls received gynaecological investigation) with no tumours found. Seven of the 10 patients were tested for thyroid peroxidase Abs. Three patients had elevated concentrations (one with GAD Abs, one with GAD Abs and positive VGKC Abs and one with low positive VGKC Abs only).

Two patients underwent temporal lobe epilepsy surgery, when the diagnosis of limbic encephalitis had not yet been made. One patient had high-titre GAD and low-titre VGKC Abs, one had no Abs. Both revealed histopathological signs of chronic lymphocytic-microglial encephalitis, as previously described in adult series.14,,17

Nine patients received corticosteroids (seven intravenously, two orally), intravenous immunoglobulins (IVIG) or combinations of both (see table 1). Median time from disease manifestation until treatment was 4 months (0–18 months) for steroids and 18 months (0.5–33 months) for IVIG. The median doses were 55 mg/kg methylprednisolone-equivalent (range 15–410 mg/kg) and 2 g/kg IVIG (range 2–2.8 g/kg).

The outcome was highly variable but largely unfavourable, with only two patients showing a good recovery according to the judgement of the physicians in charge, and there was no clear association between Ab status, treatment intensity and outcome. Both patients with onconeural Abs remained impaired. In the VGKC Ab group, three out of four patients did not recover. In the GAD Ab group, however, one of four patients recovered without being treated. Only three of 10 patients did not develop mediotemporal atrophy (two non-atrophic patients recovered clinically; in addition, the patient operated on for neuroblastoma did not lose mediotemporal tissue based on MRI). The two patients with favourable outcomes received the least intense immunotreatment (respectively, none and 15 mg/kg methylprednisolone-equivalent). In one patient, unusual rapid improvement in cognitive and behavioural dysfunction was noted after one intravenous pulse of methylprednisolone (patient 8).


Limbic encephalitis is increasingly recognised in the adult population. The majority of adult cases with non-paraneoplastic Abs are not associated with tumours5 18 and often show a good response to immunotherapies, but the disease is not often reported in children and adolescents. Here we describe limbic encephalitis in 10 children; most patients were adolescents, and nine of the 10 patients were female. PLE was diagnosed only twice, on the basis of the Ab tests, but a tumour found in only one (ganglioneuroblastoma); tumour searches are continuing in all other cases. Not all PLE patients have onconeural Abs and thorough tumour searches in patients with the clinico-neuroradiological syndrome of limbic encephalitis are recommended.1 The clinical and morphological outcome despite use of corticosteroid or IVIG treatment was disappointing. Therapeutic trials of earlier and more intense immunotherapies should be undertaken. Abs known from adult cases were found in eight of 10 paediatric patients, but their relevance in childhood cases needs further study. For instance, only one patient with VGKC Abs (in the positive range) recovered, whereas limbic encephalitis with VGKC Abs – in the high-concentration range – in adults usually has a favourable prognosis. This may be because the Abs were generally not looked for early during the children's illnesses; although high titres of onconeural Abs (eg, Hu and Ma1/2) and GAD Abs usually remain in patients even after treatment, VGKC Abs tend to show a monophasic time-course with rapid falls after treatment3 and earlier testing might have demonstrated higher Ab levels.

The VGKC Abs at concentrations in or close to the low positive range (143 and 157 pM) found in two GAD Ab positive patients are difficult to explain; these are only rarely seen together in adult cases. On the other hand, one of four GAD Ab positive patients recovered, which is unusual in view of the existing literature on this patient subgroup.5 10 The less specific thyroid peroxidase Abs were found in 3/7 patients tested. This would suggest Hashimoto encephalopathy (or, in more modern terms, steroid-responsive encephalopathy associated with autoimmune thyroiditis19), but the more specific diagnosis of limbic encephalitis surpasses this diagnosis. Thyroid Abs in limbic encephalitis have also been observed in adult onset cases.19 20

Taken together, the clinico-radiological and Ab similarities between childhood-adolescence and adulthood limbic encephalitis suggest that both result from principally identical pathogenic processes. Although not yet formally proven, it seems that the paraneoplastic encephalitides are mainly T cell mediated and the onconeural antibodies are merely markers of immune reactivity towards a tumour that expresses onconeural antigens; by contrast, in the non-paraneoplastic conditions, the antibodies to neuronal membrane receptors and ion channels are likely to be the pathogenic entity. Much greater recognition of childhood cases and systematic studies are required to ensure that these patients are diagnosed promptly and suitably treated. To facilitate this, we provide recommendations for the diagnostic work-flow and diagnostic categorisation for limbic encephalitis in figure 2 and the following explanatory text section, which applies to limbic encephalitis patients of all ages.

Figure 2

Suggested diagnostic flow-chart for limbic encephalitis. The numbers within the fields are taken up in the explanatory section at the end of the Discussion section. LE, limbic encephalitis.

  1. There are two prerequisites for the diagnosis limbic encephalitis: Signs and symptoms predominantly (but not necessarily exclusively) suggestive of limbic origin: impairment of recent memory, temporal lobe seizures or affective abnormalities for no longer than 5 years (field 1 in figure 2). Limbic encephalitis is usually a differential diagnosis in subacutely (rather than in very rapidly) evolving conditions. There may even be quite chronic forms like limbic encephalitis with GAD Abs (as patient 6 of this series). However, a hyperintense mediotemporal lesion without regression of hypersignal or volume (see below regarding field 9) over 1 to several years is unlikely to be encephalitic in origin (for examples, see the non-inflammatory group of hyperintense amygdala lesions in Soeder et al,21 supplementary table 3). To account for these time considerations, the upper limit of a period of 5 years is suggested.

  2. Hyperintense mediotemporal T2/FLAIR signal with no better explanation than inflammatory in origin (field 2). At this point, the following differential diagnoses need to be explored and excluded (for a summary from adult neurology with the results, which also apply to paediatric new-onset mediotemporal epilepsies, see Soeder et al21):

    • low grade glioma22

    • cortical dysplasia23

    • hippocampal signal increase due to status epilepticus24,,26

    • infectious encephalitis due to herpes simplex virus, human herpesvirus 627 or varicella zoster virus28

    • anti-NMDAR encephalitis (diffuse encephalopathy with no predominant mediotemporal involvement)11 29

    • chorea-akanthocytosis30 (elevated creatine kinase in serum, absence of normal chorein band on Western blot).

Mediotemporal lobe epilepsy with hippocampal sclerosis (MTLE-HS) formally fulfils the above criteria (temporal lobe seizures plus hippocampal T2/FLAIR hypersignal – here, with reduced hippocampal volume). It is questionable if MTLE-HS can be considered a true differential diagnosis to limbic encephalitis.31 MTLE-HS can result from very different types of primary brain injuries32 including limbic encephalitis.33 It is therefore recommended this diagnostic work-flow is followed in cases of recent onset MTLE-HS.31 The patient of figure 1 illustrates this: 5 years into the disease, she had right-sided hippocampal sclerosis on MRI but on histopathology there was pyramidal cell loss, astrogliosis and chronic inflammatory infiltrates. Very recently, in two large series 16% and 38% of patients with the clinical features of limbic encephalitis and well-defined Ab reactivities were reported to be MRI negative.34 35 This suggests that a mediotemporal MRI hypersignal is not always a prerequisite for the diagnosis of limbic encephalitis.

If (1) and (2) are present (figure 2), limbic encephalitis is the suspected diagnosis (field 3). Its confirmation requires at least one additional finding of the following three. (i) An Ab (field 4; a list of Abs to be searched for is given in table 2). These Abs have been found to be strongly associated with limbic encephalitis and not or only exceptionally in controls. (ii) A tumour diagnosis within 5 years of onset of limbic encephalitis symptoms (field 5). A tumour search is mandatory if limbic encephalitis is suspected. It should cover the following body compartments: thorax and abdomen, pelvis in female patients, testicles in male patients and peripheral blood. The extent of investigational modalities used may depend on the degree of suspicion for a paraneoplastic condition. If an onconeural Ab is found, if multiple levels of the nervous system are involved, or if other clinical features suggest a malignant condition, increased awareness is mandatory (as was the case with patients 1 and 2 from this series, one of whom was finally found to harbour a neuroblastoma). (iii) Histopathological evidence of mediotemporal inflammation (field 6). Whereas diagnostic mediotemporal lobe brain biopsies are rarely done, a mediotemporal hyperintense lesion not considered to be encephalitic in origin may be resected for therapeutic reasons. In such cases, encephalitis becomes evident only upon histopathological studies (as in patients 9 and 10 of this series).

Table 2

Autoantibodies relevant for the diagnostic work-up of limbic encephalitis

If none of the three qualifying features is present, the diagnosis remains ‘suspected limbic encephalitis, so far non-paraneoplastic in origin’ (field 8). Repetition of the tumour search is recommended if features suggestive of an underlying malignancy become manifest. MRI follow-up studies over 1–3 years are recommended. If they show a gradual resolution of the increased signal or of the mediotemporal volume (field 9), the diagnosis of ‘possible limbic encephalitis’ (field 10) can be made. If no such evolution occurs, it is likely that the diagnosis of limbic encephalitis needs to be reconsidered and a static pathology is the more likely diagnosis.21

It has been suggested that immunotherapy should be started as soon as limbic encephalitis is suspected.36 The ideal type of treatment has, however, not yet been established and may differ between the Ab-defined subtypes of limbic encephalitis. Many consider high-dose steroids3 or IVIG37 as viable first-line agents. Once the results of Ab diagnostics are available, these findings may contribute to further treatment decisions. Falls in titres of VGKC and NMDAR Abs (and perhaps GAD Abs) may serve as paraclinical markers for the effect of immunotherapies. Patients with Abs to membranous antigens such as VGKC and NMDARs also benefit from plasma exchange.


The authors thank Dr F Woermann, Epilepsy Centre Bielefeld-Bethel, and Professor Dr W Heindel, Department of Neuroradiology, University of Muenster, for providing MR images for the figure. They also thank Claudia Ullmann, Bonn, and Dr Leslie Jacobson, Oxford, for excellent technical assistance with antibody testing.



  • Competing interests Dr Vincent and her department in Oxford receive royalties and payments for antibody assays for the diagnosis of different neurological disorders. Dr Vincent is a consultant for Athena Diagnostics Inc (USA) and RSR Ltd (UK), and holds a patent on VGKC-antibody targets.

  • Provenance and peer review Not commissioned; externally peer reviewed.