Article Text

Research paper
Persistent anterograde amnesia following limbic encephalitis associated with antibodies to the voltage-gated potassium channel complex
  1. Christopher R Butler1,
  2. Thomas D Miller1,
  3. Manveer S Kaur2,
  4. Ian W Baker2,
  5. Georgie D Boothroyd2,
  6. Nathan A Illman3,
  7. Clive R Rosenthal1,
  8. Angela Vincent1,
  9. Camilla J Buckley1
  1. 1Nuffield Department of Clinical Neurosciences, John Radcliffe Hospital, Oxford, UK
  2. 2Russell-Cairns Unit, John Radcliffe Hospital, Oxford, UK
  3. 3Institute of Psychiatry, Kings College London, London, UK
  1. Correspondence to Dr Christopher R Butler, Nuffield Department of Clinical Neurosciences, John Radcliffe Hospital, Level 6, West Wing, Headley Way, Oxford OX3 9DU, UK; chris.butler{at}


Objective Limbic encephalitis (LE) associated with antibodies to the voltage-gated potassium channel complex (VGKC) is a potentially reversible cause of cognitive impairment. Despite the prominence of cognitive dysfunction in this syndrome, little is known about patients’ neuropsychological profile at presentation or their long-term cognitive outcome.

Methods We used a comprehensive neuropsychological test battery to evaluate cognitive function longitudinally in 19 patients with VGKC-LE.

Results Before immunotherapy, the group had significant impairment of memory, processing speed and executive function, whereas language and perceptual organisation were intact. At follow-up, cognitive impairment was restricted to the memory domain, with processing speed and executive function having returned to the normal range. Residual memory function was predicted by the antibody titre at presentation.

Conclusions The results show that, despite broad cognitive dysfunction in the acute phase, patients with VGKC-LE often make a substantial recovery with immunotherapy but may be left with permanent anterograde amnesia.

  • Cognition
  • Neuropsychology
  • Neuroimmunology
  • Limbic System
  • Amnesia

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 (LE) is a neurological syndrome characterised by acute or subacute amnesia and seizures and is frequently associated with autoantibodies against components of the voltage-gated potassium channel complex (VGKC):1 leucine-rich glioma inactivated 1 (LGI1) and contactin-associated protein-like 2 (CASPR2).2 ,3 Patients with VGKC-LE typically show clinical improvement following a course of immunosuppression.1 ,2 MRI may reveal focal abnormalities, especially T2 hyperintensity and subsequent atrophy, within the medial temporal lobes (MTL),1 ,4–7 a region critical for learning and memory.8 ,9 To date, cognitive outcome has been assessed in single cases and small case series, some of which have reported residual memory deficits1 ,10 ,11 and others apparent recovery.5 ,12 The conflicting nature of these reports is likely attributable to the small number of patients studied (nine5 and one12 in reports of normal cognitive function and seven,1 one10 and three11 in reports of persistent memory deficits) and the limited range of neuropsychological tests often used. Therefore, the extent of cognitive impairment and the degree to which VGKC-LE responds to treatment remain largely unknown. Here, we present data from a comprehensive battery of clinical neuropsychological tests administered to 19 patients before and after immunotherapy for VGKC-LE.


Patient identification

We identified 19 patients who had been diagnosed with LE (ie, had the clinical features of ‘subacute onset…of seizures, short-term memory loss, confusion, and psychiatric symptoms’13) by an experienced neurologist (CJB), had tested positive for serum VGKC-complex antibodies (AV) and had undergone neuropsychological assessment (IWB, Assessment 1) during the acute phase of the illness at the Russell Cairns Unit, Oxford, between November 1998 and May 2009. Details of the clinical presentation, investigations and response to treatment were obtained from the medical notes. All of the patients were fluent in English and had no history of previous psychiatric or neurological disorder that may result in cognitive impairment. Ethical approval for the study was granted by North Staffordshire Research Ethics Committee. All participants gave written, informed consent.


Assessment 1 was conducted before or shortly after commencement of immunotherapy. When the assessment occurred after treatment initiation (eight patients), the delay was in the order of a few days and unlikely to have allowed significant clinical improvement. Assessment 2 was conducted 3–44 months after Assessment 1 and involved the same test battery. In order to minimise practice effects, alternative forms of tests, where available, were used in Assessments 1 and 2. Patients’ scores were standardised according to published normative data. The neuropsychological battery assessed the following domains:

  • Premorbid intelligence: National Adult Reading Test14 or Wechsler Adult Reading Test.15

  • Memory: immediate and delayed recall of the Logical Memory Story (from Wechsler Memory Scale-III);16 immediate and delayed recall of a 15-item word list from the Adult Memory and Information Processing Battery (AMIPB);17 immediate and delayed recall of a complex figure from the AMIPB or the Rey–Osterreith Complex Figure.18

  • Executive function: letter and category verbal fluency;19 Trail making test part B;20 digit span from the Wechsler Adult Intelligence Scale Third Edition (WAIS-III).21

  • Processing speed: digit-symbol coding from the WAIS-III.21

  • Language: vocabulary and similarities subtests from the WAIS-III.21

  • Perceptual organisation: block design and matrix reasoning subtests from the WAIS-III;21 copy of the AMIPB17 or Rey–Osterreith Complex Figure.18

Statistical analyses

To compare performance across patients, assessment sessions and cognitive domains, raw scores from individual tests were converted to z-scores (population mean=0, SD=1). An index score for each cognitive domain was calculated for each patient by averaging their z-scores across the tests within that domain. Only the delayed recall scores were used for the memory indices as these were highly correlated with immediate-recall scores (story recall: r=0.76, p<0.001; list recall: r=0.65, p=0.004). Individual impairment on a test was defined, in line with standard neuropsychological practice, as a z-score of ≤−1.67, corresponding to the 5th centile. Group performance at each assessment was compared with the normal population distribution using one-tailed Student t tests. The change in performance between Assessments 1 and 2 was analysed using repeated measures t tests. Results were corrected for multiple comparisons using the Bonferroni method. Linear regression was used to identify predictors of cognitive outcome. Statistical analyses were performed using SPSS V.19.


Demographic and clinical features

The individual clinical data of the patients (12/19 male, mean age 60.1±14.6 years) are shown in table 1. The mean estimated premorbid IQ was 104±9.6. The median delay from symptom onset (when the patient or relative first noticed symptoms potentially attributable to VGKC-LE) to treatment was 144 days (range 30–408 days). Often the long delay was because they first presented at a time when VGKC-LE had only recently been recognised. Hyperintensity on T2-weighted and fluid-attenuated inversion recovery MRI sequences was seen in the MTL in 9/11 patients from whom imaging was available for review. The median VGKC-complex antibody titre at Assessment 1 was 925 pmol/L (range 177–6000 pmol/L). Of three patients with titres <400 pmol/L, two (patients 11 and 13 in table 1) developed titres >400 pmol/L upon repeat testing. A total of 9/10 patients with serum available for later analysis had LGI1 antibodies and only one had CASPR2 antibodies.

Table 1

Clinical features and individual verbal memory scores at Assessments 1 and 2 for the 19 VGKC-LE patients

Cognitive presentation of VGKC-LE (Assessment 1)

Assessment 1 was conducted at a median of 111 days (range 16–377 days) after the onset of symptoms. Group results for the cognitive domain index scores are shown in table 2. The results for individual tests can be found in online supplementary table S1.

Table 2

Group performance on cognitive domain indices at Assessment 1 (acute phase) and Assessment 2 (post-treatment phase) of VGKC-LE

Memory: Group performance revealed significant impairment on the index of verbal memory (mean=−1.76±1.02; t=−7.47, df=17, p<0.001; figure 1A). In all, 12/18 patients (one patient did not complete both assessment sessions) fell below the 5th centile. In particular, the group was impaired on story immediate recall (mean z=−0.85±1.00, t=−3.61, df=17, p=0.001) and delayed recall (mean z=−1.75±1.39, t=−7.42, df=17, p<0.001), and word list learning (mean z=−1.63±0.57, t=−6.31, df=14, p<0.001) and delayed recall (mean z=−1.76±0.99, t=−6.82, df=14, p<0.001). The greatest proportion of patients was impaired on delayed recall of the word list (10/15 patients). The group performance on the non-verbal memory index was not significantly impaired when corrected for multiple comparisons (mean=−0.52±1.21; t=−2.14, df=16, p=0.024).

Figure 1

Box-and-whisker plot showing the six cognitive domain indices (A) before and (B) after treatment. Individual data points for each patient are plotted on top of each index box-and-whisker plot. The line in the box plot indicates the median value. Prior to treatment there were impairments in verbal memory, executive function and processing speed, when compared with normative data (denoted by, * where p<0.008 (Bonferroni-corrected significance level)). Following treatment, the only persistent deficit was in verbal memory. The hatched line represents a z-score of −1.67 corresponding to the 5th centile. (C) The correlation between VGKC-complex antibody (VGKC-complex Ab) titre at presentation and score on the verbal memory index at Assessment 2.

Executive function: Group performance revealed significant impairment on the index of executive function (mean=−0.90±1.00; t=−3.92, df=18, p<0.001). The number of patients in the impaired range was only 1/19. The group as a whole was impaired on letter fluency (mean z=−0.94±0.95, t=−4.01, df=17, p<0.001), category fluency (mean z=−0.62±1.58, t=−2.71, df=18, p=0.007) and Trails B (mean z=−1.64±2.59, t=−6.35, df=14, p<0.001).

Processing speed: Performance on the digit-symbol coding task was reduced across the group (mean z=−0.76±0.80, t=−3.13, df=16, p=0.003), with one patient performing in the impaired range.

Language: There was no significant impairment on the language index (mean z=0.20±0.86, t=0.89, df=18, p=0.808).

Perceptual organisation: There was no significant impairment on the perceptual organisation index (mean z=0.28±0.49, t=1.22, df=18, p=0.881).

Cognitive outcome of VGKC-LE (Assessment 2)

The median interval between Assessments 1 and 2 was 254 days (range 83–1395 days). The median antibody titre at Assessment 2 was 116 pmol/L (range 0–585 pmol/L) (see table 1), significantly reduced from the median titre at Assessment 1 (Wilcoxon Signed Rank test: p<0.001). Group results for the cognitive domain index scores are shown in table 2. The results for individual tests can be found in online supplementary table S1.

Assessment 2 (figure 1B) revealed a persistent group level impairment in the index of verbal memory (mean=−1.07±0.96, t=−4.54, p<0.001). Overall, 5/19 patients still fell below the 5th centile. Specific impairments were observed in delayed recall of the story (mean z=−0.93±0.88, t=−3.95, p<0.001) and word list (mean z=−1.37±1.02, t=−5.65, p<0.001). There was no impairment on the indices or any individual test of non-verbal memory, executive function, processing speed, language or perceptual organisation.

Cognitive improvement in VGKC-LE (Assessment 1 vs Assessment 2)

Between Assessments 1 and 2, there were significant improvements in the index scores for verbal memory (mean difference=0.84±1.05, df=17, t=3.30, p=0.005) and executive function (mean difference=0.88±0.76, df=18, t=5.03, p<0.001; see table 2). Improvements in performance on individual tests can be seen in online supplementary table S1.

Predictors of cognitive outcome

Multiple regression with simultaneous variable selection was used to identify predictors of cognitive outcome. The verbal memory index at Assessment 2 was entered as the dependent variable as it was considered most likely to be sensitive to MTL pathology and was the only index demonstrating long-term impairment. The independent variables entered into the model were: age; delay from symptom onset to initiation of treatment; interval between symptom onset and Assessment 2; and VGKC antibody titre at presentation (log transformed to ensure normal distribution; figure 1C). Only variables that met the forward selection criterion (p<0.05) were retained. Antibody titre at presentation significantly predicted the verbal memory index value at Assessment 2 (β=−0.54, p=0.039). No other independent variables were retained.


This study is the first to investigate the cognitive outcomes of VGKC-LE using comprehensive, longitudinal neuropsychological assessment in a large group of patients. Our results demonstrate that, at presentation, patients showed marked impairment on a range of measures of memory, executive function and processing speed, whereas language and perceptual organisation were unaffected. This pattern is consistent with acute dysfunction of the limbic system and is similar to that seen in acute viral LE.22

Following immunotherapy, residual deficits were observed on verbal memory measures, with about a third of patients remaining in the impaired range (<5th centile) on delayed word list recall, despite the recovery of executive function and processing speed. The VGKC-complex antibody titre at presentation predicted the extent of these residual verbal memory deficits.

Our results suggest that VGKC-LE leads to a more focal cognitive impairment than other forms of autoimmune encephalitis, such as that associated with antibodies to the N-methyl-D-aspartate (NMDA) receptor.23 It is important to note that, from these data alone, it is not possible to be certain that the patients’ cognitive improvement was a direct result of treatment, since we do not have a ‘no treatment’ control group. The persistent memory deficits following VGKC-LE are potentially attributable to focal hippocampal atrophy that is often reported after recovery from the acute phase,4 ,7 ,11 ,24 and which may be more common in patients with very high antibody titres.6

These results are important as they identify cognitive markers of disease outcome that can be used both by clinicians and in future research. They are also of relevance to the cognitive neuroscience community. The inclusion of patients with treated VGKC-LE in influential studies of memory25 has been criticised26 on the grounds that deficits might extend beyond the memory domain. Our results provide evidence that, at the group as well as individual patient level, cognitive impairment following VGKC-LE is restricted to amnesia. Further work in larger patient groups with systematic neuroimaging is needed to investigate potential causal relationships among antibody titre, treatment, hippocampal atrophy and cognitive outcomes.


Our thanks go to the patients and relatives who participated in this research and to the consultant colleagues who referred patients.


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:


  • Contributors CJB, IWB, MSK and AV originally conceived the study. CRB, TDM, MSK, IWB, GDB, NAI and CRR collected and analysed the data. All authors contributed to the interpretation of results and writing of the paper. CRB, AV and CJB are the guarantors.

  • Funding This study was partially funded by a research prize from the Encephalitis Society (CJB, MSK, AV, IWB).

  • Competing interests AV and the Department of Clinical Neurology in Oxford receive royalties and payments for Ab assays. AV is the inventor on patent application WO/2010/046716 entitled ‘Neurological Autoimmune Disorders.’ The patent has been licensed to Euroimmun AG for the development of assays for LGi1 and other VGKC-complex Abs. AV has done paid consultancy for Athena Diagnostics and is employed by Oxford University. CRB is funded by a Clinician Scientist Fellowship from the Medical Research Council. TDM is supported by the Encephalitis Society and Patrick Berthoud Charitable Trust. CJB was funded by a Clinician Scientist Fellowship from the Medical Research Council. CRR is funded by National Institute for Health Research (NIHR) Oxford Biomedical Research Centre based at Oxford University Hospitals NHS Trust and University of Oxford, which also supports the serological work. The funding sources had no involvement in the collection, analysis or interpretation of data.

  • Ethics approval North Staffordshire Research Ethics Committee.

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