Statistics from Altmetric.com
Non-infective, antibody-negative limbic encephalitis presents a diagnostic and therapeutic challenge. The treatment aims to minimise the functionally devastating and irreversible cognitive impairment. An illustrative case history and review of the literature are presented and a management algorithm proposed.
Limbic encephalitis (LE) typically presents with amnesia, behavioural disturbance, psychiatric symptoms, seizures and altered consciousness. Although the aetiology was historically considered paraneoplastic, LE may also result from autoimmune processes, independent of malignancy. New central nervous system targets are being identified and our appreciation of the associated clinical phenotypes is evolving rapidly. With this improved understanding, specific treatment protocols are emerging. Despite these developments, patients with LE without the characteristic antibodies still present a considerable clinical challenge. We present a case history and review the treatment and management options for antibody-negative LE.
A 65-year-old female personal assistant presented with 3 months of confusion and disorientation. Her family reported that she showed increasing depression, general malaise and fatigue. Her work performance had deteriorated as she had become increasingly forgetful, repetitive, anxious and irritable. Treatment in the community for a urinary tract infection had no discernible benefit. She had previously been treated for postnatal depression, but had been otherwise well. Her friends and relatives described her premorbid personality as cheerful, health-conscious, hard-working and well organised with good memory and social skills. She was a non-smoker and she drank one bottle of wine a week.
On examination, she appeared anxious and had a coarse resting tremor. The reflexes were brisk but the plantars were flexor. The Addenbrookes' cognitive examination (ACE-R), although difficult to perform early in her presentation, was 69/100, with normal fluency and language, but poor attention and orientation to time, person and place. There was profound impairment of episodic and personal memory.
Serum and urine osmolalities were consistent with the syndrome of inappropriate antidiuretic hormone secretion. Other blood tests were normal, including full blood count, renal and liver function, inflammatory markers (C-reactive protein and plasma viscosity), angiotensin-converting enzyme, serum protein electrophoresis, serum B12 and folate. Serology for herpes simplex virus (HSV), human herpes virus 6, hepatitis B and C, JC virus, Borrelia burgdorferi, syphilis, toxoplasma and human immunodeficiency virus were negative. Serum IgG (but not IgM) for Epstein–Barr virus and cytomegalovirus were positive, consistent with past exposure. The autoimmune profile was similarly uninformative, including negative testing for antinuclear antibody, double-stranded DNA, anti-thyroid peroxidase antibody, anticardiolipin antibody and antineutrophil cytoplasmic antibody.
MRI of the brain showed high signal changes involving the medial temporal lobes and hippocampi bilaterally, without enhancement (figure 1). There were also high signal changes in the right medial thalamus. Electroencephalography (EEG) several days after admission showed non-specific bi-temporal slow activity. Initial cerebrospinal fluid (CSF) analysis showed a white blood cell count of 15/mm3 (100% lymphocytes), protein 0.52 g/l, glucose 3.9 mmol/l (matched plasma glucose 9.0 mmol/l), lactate 1.6 mmol/l and unmatched oligoclonal bands. CSF culture showed no bacterial growth, including prolonged culture for mycobacteria. PCR for HSV, enterovirus, Varicella zoster, cryptococcus and cytomegalovirus was negative. There were no CSF antibodies to voltage-gated potassium channels (VGKC) and α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptor.
CT of the chest, abdomen and pelvis, mammogram and whole body fluorodeoxyglucose-positron emission tomography (PET) showed no evidence of malignancy. Extended autoimmune screening for antibodies associated with LE, including anti-N-methyl-D-aspartate (NMDA) receptor antibody, VGKC antibody, AMPA receptor antibody, leucine-rich glioma-inactivated-1 (LGI-1) and glutamic acid decarboxylase antibodies, was also negative. A paraneoplastic screen (anti-Yo, anti-Hu, anti-Ri, amphiphysin, Ma1/2, collapsin response mediator protein-5/crossveinless-2 antibodies) was also negative.
The differential diagnosis at presentation included herpes simplex encephalitis and we prescribed intravenous aciclovir and thiamine. Subsequently, she completed a course of intravenous methylprednisolone (1 g daily for 3 days), followed by five sessions of plasma exchange. She started oral prednisolone (60 mg/day), with the dose gradually withdrawn over 6 months. We introduced citalopram 3 weeks after admission, primarily as an anxiolytic.
In the weeks following plasma exchange, both family and hospital staff noted subjective improvement; her anxiety was less pronounced and she showed limited acquisition of new memories—for example, for navigation around the ward and for other patients' names. The ACE-R improved (from 69 to 75) and her scores on tests of attention and recall increased (table 1). More detailed cognitive assessment became possible, although still limited. Her episodic memory remained significantly impaired but with intact semantic memory; her visual and verbal memory were both delayed and there was mild impairment of attention, visuospatial and constructional skills. Repeat MRI brain 4 weeks after starting treatment showed improvement in the appearance of the medial temporal lobes, particularly on the right, but with residual high signal in the hippocampi and right medial thalamus (figure 1).
Her ACE-R score had improved further at 4 months after plasma exchange, particularly in the memory domain (total score 84; table 1). Detailed neuropsychological assessment 5 months after presentation (table 2) showed well-preserved general intellect and language but, with persistent severe anterograde amnesia. Executive test scores were variable, and some scores fell within the impaired range. Her functional status was consistent with these results, characterised by a severe deficit in explicit learning, and, to a lesser extent, in implicit learning. She showed a temporally graded retrograde amnesia, able to recall more recent, pre-illness information, but appearing quite disorganised. Executive difficulties impaired her ability to plan and organise effectively. Her anxiety was less pronounced. On repeat MRI brain, the high signal was limited to the now atrophic hippocampi (figure 1).
This patient presented with a subacute amnesic syndrome with radiological changes consistent with LE, but without any identified autoantibody. Although we cannot confidently exclude a paraneoplastic cause until after at least 5 years without the emergence of an underlying malignancy,1 2thorough investigation has not, to date, identified a tumour. The pointers towards an autoimmune inflammatory aetiology include the negative screening for antibody-mediated LE, the presence of oligoclonal bands and the improvement following immunotherapy.
LE involves the limbic system and mesio-temporal lobes. Our patient illustrated this, both in her constellation of symptoms and signs (cognitive impairment, neuropsychiatric features and seizures) and in the distribution of signal change on MRI and PET imaging.2,–,4 CSF studies in LE may show inflammatory changes with lymphocytic pleocytosis, raised protein and positive oligoclonal bands.2,–,5 EEG may show non-specific generalised or focal slowing, occasionally with epileptiform activity and clinical seizures.2 4 5 Several antibodies have been convincingly associated with immune-mediated LE and this remains a focus of active research. Some of the newly identified antibodies cannot be identified by conventional immunohistochemistry. These antibodies can be broadly classified into cytotoxic T cell mediated antibodies against intracellular neuronal antigens, and antibodies against neuronal cell membrane antigens (table 3).2 5
The diagnosis of LE presents a dilemma for neurologists (table 4), The clinical and radiological features do not reliably differentiate paraneoplastic from non-paraneoplastic LE, and diagnostic antibody tests are often delayed or negative. However, some features are more commonly associated with one or other autoantibody. Patients with LE secondary to malignancy, as well as characteristic LE features, may have constitutional symptoms and other neurological paraneoplastic syndromes. NMDA receptor-associated LE may present with seizures, autonomic dysfunction, psychiatric symptoms and dyskinesias; its well-described association with ovarian teratoma in young women must not be overlooked.1 2 5 VGKC antibody-positive LE typically presents with hyponatraemia, autonomic symptoms and seizures, in addition to amnesia.2 5 6 It is infrequently associated with cancer and has a reasonably favourable prognosis.4 5 Approximately 30% of patients with VGKC antibody-associated encephalitis have an underlying tumour, mostly small cell lung cancers and thymoma.2 7
Although there are protocols for treating antibody-associated LE,4 6 8 these are based on level 4 evidence, with no randomised controlled trial evidence. In general, however, patients with positive neuronal surface antigen antibodies improve after immunotherapy.4,–,8 Anecdotal evidence suggests that those with paraneoplastic disease may respond to tumour resection, chemotherapy or/and radiotherapy as well as immunotherapy, although the prognosis remains guarded.3 5 Patients with VGKC antibodies are normally recommended to have intravenous Ig or plasma exchange at presentation for rapid antibody clearance, followed by at least 6-months' treatment with high-dose corticosteroid.4 Antibody titres tend to correlate with the clinical response in seropositive LE: normalisation of antibody titres (<100 pM) in VGKC antibody-positive LE is associated with cognitive improvement.4 8 Some patients relapse after corticosteroid withdrawal, necessitating further treatment with intravenous Ig or an increased corticosteroid dose. Improvement is measured mainly by improvement in cognitive assessment scores, or improvement in seizure activity. A raised titre without further clinical relapse is not itself an indication for more immunosuppression. However, persistent cognitive impairment or seizure with persistent raised antibody would warrant further immunotherapy.
Patients treated early with corticosteroids often show clinical improvement and reduced seizure frequency, whereas those for whom corticosteroids are delayed show variable improvements and slower falls in VGKC antibody levels.4 6 Starting corticosteroids within 2 months of symptom onset is generally associated with a more favourable outcome.6 Cortical atrophy correlates with high initial serum VGKC antibody levels and tends to be associated with persistent cognitive deficits, especially on verbal and visual memory.3 8 Wong et al8 used a standardised immunotherapy regimen in patients with VGKC antibody-positive LE, and noted a ‘time window’ when clinical improvements could be expected; they saw seizure remission and resolution of hyponatraemia within 1 week of treatment, improved memory function within 1–3 months, normalised VGKC antibody titre within 1–4 months and improved hippocampal swelling on MRI at 6–9 months. There are few data to support using corticosteroid-sparing agents: cyclophosphamide, azathioprine, mycophenolate mofetil and rituximab each show only limited efficacy.2,–,5 8
Patients with LE and where characteristic antibodies cannot be detected are a challenging subgroup. Empirical use of corticosteroids or other immunotherapies remains controversial, although anecdotal evidence supports using corticosteroid treatment in the initial stages of the illness; some patients require no further treatment.4 9 10 However, the options for further additional immunotherapy include intravenous Ig (0.4 g/kg/day for 5 days), followed by 50 mg/day oral prednisolone, gradually withdrawn over 6 months.4 10 Suggested alternatives include plasma exchange and monthly pulsed intravenous methylprednisolone (1 g daily for 3 days).11
Treatment recommendations for antibody-negative LE
Having established the diagnosis of antibody-negative LE, including undertaking a screen for malignancy, the clinician must consider the relative risk and benefits of immunotherapy. We propose a treatment algorithm for antibody-negative, non-infective LE, extrapolating from treatment regimens reported to be of benefit in seropositive and seronegative LE,3 4 6,–,11 (figure 2). There is level 3 evidence from case series for initial treatment with intravenous methylprednisolone (1 g daily for 3 days), followed by high-dose oral prednisolone (1 mg/kg/day). The next step, unless corticosteroids give clear benefit, is to consider plasma exchange (over five sessions) or intravenous Ig (0.4 g/kg/day for 5 days), although the benefit may be limited and the risk–benefit ratio should be carefully assessed. Whenever possible in patients with antibody-negative LE, serological testing for antibodies against new cell-membrane antigens, undetectable by conventional immunohistochemistry, should be tested in research laboratories (eg, antibodies to the neuropil of the hippocampus and cerebellum, GLuR1/2 subunits of AMPA receptor and VGKC antibodies directed against hippocampal protein LGI-1).
Serial MRI brain scans may provide additional evidence of improvement, with reduced signal change with treatment. Furthermore, atrophy of the hippocampus, temporal lobe and amygdala may predict persistent cognitive impairment.4 8 9 and, in the absence of a biochemical marker, clinical progress should be monitored with repeat cognitive assessments. We propose repeat follow-up cognitive assessment by a neuropsychiatrist or neuropsychologist at 1, 3, 6 and 12 months after treatment. These intervals permit treatment adjustment and repeat imaging if necessary. Oral prednisolone should be tapered gradually over 4–6 months (extrapolating from VGKC antibody-positive case series).4 8 Treatment options for those who are unresponsive to corticosteroids or other immunotherapy remain unclear. Cortical destruction and atrophy suggest a poor prognosis. It is important to continue the investigation and monitoring for an underlying neoplastic process for at least 5 years.1 3 Neurorehabilitation and multidisciplinary input are essential, particularly in patients with continued cognitive impairment.
Treating patients with non-paraneoplastic, antibody-negative LE is particularly challenging. With the advent of new immunological techniques, further antibodies are being identified which will enable improved characterisation of these syndromes and better monitoring of the response to treatment. Although not proved, it is intuitive that those with seronegative LE should be treated early to minimise functionally devastating and irreversible cognitive impairment. Our proposed follow-up regimen should improve monitoring of treatment response, allow timely detection of relapse and enable escalation of immunotherapy if required.
The authors are grateful to the patient and her family for permission to publish this case report. They acknowledge the expertise of Professor Angela Vincent and her team at the Department of Immunology, John Radcliffe Hospital, Oxford and greatly appreciate their assistance in performing the extended immunology screen.
Funding CMR is funded by the National Institute for Health Research and The Burden Institute.
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed by Jeremy Rees, London, UK.
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.