DIFFERENTIAL DIAGNOSIS

There are many causes of (sub)acute encephalopathy³ including:

• metabolic disorders such as uraemia and hepatic failure

• drugs including chemotherapeutic agents

• toxins including alcohol

• deficiency states including the Wernicke-Korsakoff syndrome

• inflammatory disorders including acute disseminated encephalomyelitis

• primary or secondary central nervous system (CNS) malignancy including lymphoma

• neurodegenerative disorders including Creutzfeldt-Jacob disease.

Herpes simplex is not only the commonest identified cause of viral encephalitis in general, but by far the commonest cause of viral limbic encephalitis in particular

However, in patients with symptoms mainly referable to limbic lobe dysfunction, infective and immune mediated causes are the main diagnostic considerations. Of the infections, herpes viruses, and particularly herpes simplex, are the most important. In addition, an expanding range of immune mediated causes, including some connective tissue diseases, paraneoplastic limbic encephalitis, and voltage gated potassium channel associated encephalopathy appear to be particularly associated with this clinical phenotype.

INFECTIOUS CAUSES

Infections must always be considered first and empirical treatment should not be delayed if there is any doubt about the diagnosis. Although a wide range of viral, bacterial, and tropical infections, and even neurosyphilis,⁴ have been reported in the context of a limbic encephalitis phenotype, herpes simplex is not only the commonest identified cause of viral encephalitis in general, but by far the commonest cause of viral limbic encephalitis in particular. In the immunocompetent host with viral encephalitis, 70% of cases are caused by herpes simplex type 1, but in immunocompromised patients with limbic encephalitis, herpes simplex type 2, and human herpes viruses (HHV) 6 and 7 are important possibilities.

Patients with herpes simplex encephalitis typically present with a fairly abrupt onset of confusion, memory impairment, and often seizures.⁵ Fever is common but not invariable. Neuroimaging usually reveals signal change and swelling within the temporal lobes; this is often visible on CT, but is more easily seen using MRI (fig 1). Significant brain swelling may lead to raised intracranial pressure, sometimes requiring medical or even surgical treatment.⁵ Cerebrospinal fluid (CSF) examination commonly reveals a lymphocytosis and raised protein, but the gold standard for in vivo diagnosis is CSF polymerase chain reaction (PCR) for herpes viruses, which has a sensitivity and specificity of ~95%.⁶ Untreated, there is a case fatality of ~70%,⁷ and so treatment with intravenous aciclovir, which reduces case fatality to 20–30% and has relatively few adverse effects (renal impairment and confusion), should not be delayed.⁵ Treatment in proven cases should continue for at least 14 days, longer in immunocompromised patients. Some authors recommend a repeat CSF PCR examination at the end of treatment to ensure viral clearance, and further treatment with aciclovir if continued infection is suspected, although there is limited evidence to support this approach.⁸ An often more difficult decision is when to stop treatment if the initial CSF PCR is negative. In this instance, in acutely ill patients, it may be appropriate to send a second CSF sample for PCR, while continuing treatment for at least 14 days. Clearly in such instances, investigation for alternative causes becomes more pressing.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Axial FLAIR MRI in herpes simplex encephalitis shows extensive signal change in the right temporal lobe (arrow).

In immunocompromised patients, and particularly those with HIV infection or who have undergone stem cell transplant,⁹ HHV6 infection or reactivation is an important consideration. Such patients present with fairly abrupt onset of memory impairment, sleep disturbance, and intermittent confusion, have MRI signal change that may be indistinguishable from other causes of limbic encephalitis, often with striking selective hippocampal involvement (fig 2), temporal lobe seizures, and sometimes an acellular CSF. Although the exact role of HHV6 in central nervous system diseases remains controversial, in this clinical scenario detection of HHV6 in the CSF by PCR should prompt treatment with ganciclovir and/or foscarnet,¹⁰ as there is some evidence to suggest that this may prevent hippocampal damage. Cases of HHV7 related limbic encephalitis appear to be much rarer, and are treated with foscarnet.¹⁰

• Download figure
• Open in new tab
• Download powerpoint

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Coronal (A) and axial (B) FLAIR MRI in an immunocompromised patient with HHV6 limbic encephalitis showing bilateral hippocampal signal change. (arrows).

CONNECTIVE TISSUE DISORDERS AND VASCULITIS

A number of connective tissue disorders, including systemic lupus erythematosus, Behçet’s disease, Sjögren’s syndrome, and relapsing polychondritis may present with subacute memory impairment, a range of psychiatric manifestations, and seizures. A similar syndrome may also be seen in primary central nervous system (CNS) vasculitis and other granulomatous disorders including sarcoidosis. Although in most cases there are additional CNS and systemic manifestations, there are rare reports of patients presenting with a fairly pure limbic encephalitis phenotype and even with selective MRI temporal lobe signal change.^11–14 Diagnosing these conditions is often then very challenging:

• A careful history and examination may reveal clues such as a rash, arthralgia, sicca symptoms, pathergy, or oro-genital ulceration.

• Investigations should include inflammatory markers, testing for the relevant autoantibodies in peripheral blood, imaging and CSF examination.

• Brain biopsy may ultimately be required for definitive diagnosis, and may be the only means of diagnosing isolated CNS vasculitis.¹⁵ In a series of non-targeted right frontal lobe brain biopsies to investigate patients with atypical dementia in whom other investigations had failed to provide a diagnosis, an inflammatory cause was uncovered in 9% of patients.¹⁶ If confirmed, the mainstay of treatment for these inflammatory conditions is immunosuppression.¹⁵

PARANEOPLASTIC LIMBIC ENCEPHALITIS

The association between a subacute encephalopathy and a distant tumour was first described by Brierley et al.¹⁷ Corsellis et al reported more cases in 1968, and coined the term paraneoplastic limbic encephalitis (PLE).¹ A number of different tumours and associated antibodies are now recognised to be associated with this syndrome (table 1). The pathological changes within limbic structures include perivascular lymphocytic infiltration, neuronal cell loss, and reactive microglial proliferation.¹⁸

A variety of different criteria have been used for the diagnosis of PLE. The most recent consensus for a definite diagnosis of “classical” PLE requires:

• an appropriate clinical phenotype developing over a maximum of 12 weeks (although proven cases with longer courses have been described);

• neuropathological or neuroradiological (MRI/PET/SPECT) evidence of involvement of the limbic system;

• and either a cancer discovered within five years following onset of PLE, or the presence of a well characterised onconeural antibody (table 1).¹⁹

The characteristics of patients presenting with PLE reported in two retrospective series 2, ²⁰ are summarised in table 2.

Although there is no consistent relation between phenotype and underlying malignancy,²¹ patients with anti-Hu antibodies commonly appear to have symptoms attributable to dysfunction outside the limbic system²; and patients with anti-Ma2 antibodies appear to have more frequent hypothalamic and brainstem involvement and abnormal MRI findings compared with other patients with PLE^{.2, 21} Importantly, serum antineuronal antibodies are not detected in ~40% of cases of proven PLE, and thus their absence does not exclude the diagnosis. Not all patients have temporal lobe MRI signal change; there is some evidence to suggest that PET imaging may be useful in demonstrating temporal lobe abnormalities in MRI negative cases.²²

In the largest published series to date, PLE preceded the diagnosis of cancer in 60% of cases by an average of 3½ months, and when a tumour was identified, there was evidence neither of distant nor local spread in 75% of cases.² This has implications for treatment, because not only is there a better prospect for curative therapy of the underlying tumour if it is detected early and has not metastasied, but there is also evidence that, as with other paraneoplastic syndromes,²³ treatment of the underlying tumour rather than immunosuppression may lead to a better neurological outcome.²

Traditionally, searching for an underlying tumour has been by detailed imaging of chest, abdomen, and pelvis using high resolution CT, supplemented with mammography, testicular ultrasound, and tumour markers where appropriate. An alternative strategy is to use whole body fluoro-deoxyglucose positron emission tomography (FDG-PET) imaging, which is used routinely in several UK centres to stage established malignancy. Although few prospective data are available, in a large retrospective study specifically addressing this issue, FDG-PET detected a tumour in 37% of patients with suspected PLE in whom routine imaging was normal; the false positive rate was 10%.²⁴ These and other results have led some authors to advocate FDG-PET as the primary investigative tool in suspected cases of PLE, and that this may be a cost effective strategy. There is also evidence to suggest that the combination of CT and PET may improve sensitivity and specificity.²⁵ In practice, local policies, expertise, availability of imaging, and cost considerations will determine which techniques are preferred in an individual hospital.

The mechanism by which distant malignancies cause limbic encephalitis is not clear. Although is seems likely that PLE is immune mediated, antineuronal antibodies may only be markers of cell mediated immunopathology, rather

VOLTAGE GATED POTASSIUM CHANNEL ASSOCIATED LIMBIC ENCEPHALOPATHY

Voltage gated potassium channels (VGKC) are a diverse group of membrane bound proteins responsible for repolarising the nerve terminal after the passage of each action potential. One family of such channels, Shaker VGKC (Kv1), can assemble in a variety of combinations to form a wide diversity of different channels which are expressed in different parts of the nervous system, with Kv1.1 and Kv1.2 being strongly expressed in the molecular layer of the hippocampus.

Antibodies to VGKC have been implicated in a number of different neurological conditions (see below), and they may be detected and quantified in serum or CSF by radioimmunoprecipitation using ¹²⁵I-labelled α-dendrotoxin extracted from the green mamba snake (Dendroaspis angusticeps) (fig 3) which preferentially blocks Kv1.1, Kv1.2 and to a lesser extent Kv1.6 channels.²⁹ Using this technique, the reference range in controls has been quoted as <100 picomolar (pM).²⁹

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

The green mamba snake (Dendroaspis angusticeps), source of dendrotoxin, used for the assay of certain VGKC antibodies.

Raised levels of VGKC antibodies are associated with a variety of acquired peripheral nerve hyperexcitability syndromes including cramp fasciculation syndrome, Isaac’s syndrome (acquired neuromyotonia), and Morvan’s syndrome, which comprises acquired neuromyotonia together with various CNS abnormalities including sleep disorders, autonomic dysfunction, and cognitive impairment.³⁰ The similarity of the cognitive features associated with Morvan’s syndrome with those seen with apparently non-infective limbic encephalitis prompted investigators from Oxford to measure VGKC antibodies. Their first report described two patients with limbic encephalitis in whom an infective cause had been excluded, both of whom had raised VGKC antibodies and one of whom had temporal lobe signal change on MRI.³¹ One patient had a thymoma, the other had no detectable malignancy. The latter patient improved spontaneously with a parallel fall in VGKC level, and the first patient improved markedly, following plasma exchange, again with a decline in antibody level. Following further reports of treatment responsive, apparently non-paraneoplastic limbic encephalitis associated with raised levels of VGKC antibodies,^32,33 a series of 10 patients was published by Vincent et al,²⁹ shortly followed by a series of seven patients from the Mayo clinic ³⁴; all of these 17 patients had VGKC titres >400 pM.

Although the spectrum of clinical features of VGKC associated limbic encephalitis continues to be defined, a number of core features have emerged. Patients usually present in middle age with subacute memory impairment, and a range of psychiatric features including confusion, disorientation, and behavioural change attributable to limbic dysfunction. Seizures occur in the majority, and may be very difficult to control. Neuropsychological testing, where possible, reveals fronto-temporal dysfunction with prominent episodic memory impairment and relative sparing of parietal lobe function. Hyponatraemia due to the syndrome of inappropriate antidiuretic hormone secretion (SIADH) appears to be common, precedes treatment with anti-epileptic drugs, and may itself be resistant to treatment. Most patients have EEG abnormalities including diffuse slowing or focal, usually temporal lobe sharp waves, and MRI signal change in the temporal lobes (fig 4A). Most cases are not associated with occult neoplasia, and CSF examination findings are non-specific, showing at most a mild lymphocytosis and raised protein. Rare demonstration of matched oligoclonal bands, and VGKC antibodies in CSF and serum, both disappearing in convalescence, is suggestive of inflammation arising outside the central nervous system.

• Download figure
• Open in new tab
• Download powerpoint

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Coronal FLAIR MRI in a patient with VGKC antibody related limbic encephalitis before (A) and after (B) treatment. Medial temporal signal change (arrows) reduced after treatment with the development of medial temporal lobe atrophy (arrows). than pathogenic per se.^26–28 There is some evidence to suggest that patients with certain paraneoplastic syndromes have a better prognosis from their underlying tumour than those without. This is presumed to occur as a result of an immune response directed against the primary neoplasm, the paraneoplastic syndrome resulting from cross reaction with common epitopes expressed within the nervous system.²⁷

Following treatment with varying combinations of plasma exchange, intravenous immunoglobulin (IVIg), and high dose oral steroids, most patients show a decline in VGKC antibody levels with parallel improvement in neuropsychology and seizure control. The steroids can be slowly tailed off over months, titrating against clinical state and VGKC antibody titre. Close attention to seizure control and careful monitoring of electrolytes including sodium should be continued. Seizures and hyponatraemia, which may initially be intractable, appear to remit following immunosuppression. MR signal change generally resolves, but medial temporal atrophy often remains (fig 4B), presumably explaining any persisting and at times profound cognitive deficits. Patients appear to do best if promptly treated, and there is evidence to suggest that maximum improvement is seen in those patients in whom maximum reduction in VGKC level is obtained and brain atrophy is therefore prevented.

More patients with non-paraneoplastic VGKC associated limbic encephalitis have subsequently been reported, with over 100 significantly raised antibody titres being detected in the UK (A Vincent, personal communication). Further CNS features associated with this condition continue to be described, and to date these include reversible REM sleep disorder³⁵ and steroid responsive pilomotor seizures.³⁶ A recent report suggested that although patients with high titres of VGKC antibodies may present with limbic encephalitis and seizures, lower titres may be found in patients with more chronic seizure disorders extending over several years.³⁷

There is much evidence to suggest that in VGKC associated limbic encephalitis the antibodies are pathogenic. The target antigens (Kv1.1 and 1.2) are widely expressed within the limbic system, where they are located on the plasma membrane.³⁸ In vitro, VGKC antibodies bind to VGKCs located on the molecular layer of the hippocampus.³¹ VGKC antibodies are found in CSF and serum, and there is a temporal relationship between treatment, reduction in antibody levels, and clinical improvement.²⁹ Finally, mutations in Kv1.1 channels are associated with intractable seizures both in patients³⁹ and experimental animals.⁴⁰

Further prospective studies are required to establish the epidemiology, precipitants, and range of phenotypic presentations associated with VGKC antibody related limbic encephalitis. Available evidence based on the small numbers of patients studied so far suggests that acute treatment with either plasma exchange or IVIg followed by high dose oral steroids may be the treatment of choice, but there are no randomised trials.

Raised levels of VGKC antibodies are associated with a variety of acquired peripheral nerve hyperexcitability syndromes including cramp fasciculation syndrome, Isaac’s syndrome (acquired neuromyotonia), and Morvan’s syndrome

HASHIMOTO’S ENCEPHALOPATHY

Although many consider the phenotype of “Hashimoto’s encephalopathy” to be distinct from limbic encephalitis, in that the former is often associated with more widespread MRI signal change and a wider and often different range of clinical manifestations, there are sufficient similarities and possible overlap between these two entities to warrant discussion here.

Brain et al were the first to describe a patient with encephalopathy and raised thyroid antibodies.⁴¹ Their patient, a 49 year old man with histologically confirmed Hashimoto thyroiditis, hypothyroidism, and raised serum antithyroid microsomal and antithyroglobulin antibodies, presented with a slowly progressive syndrome with impaired consciousness, episodic confusion often preceded by agitation, hallucinations and tremor, recurrent stroke-like episodes, and cognitive impairment. He did not improve following three months of prednisolone and anticoagulation, but his symptoms remitted on thyroxine alone. With reference to the raised thyroid antibodies, Brain et al concluded that “Antibody studies in future cases of unexplained encephalopathy should show whether we have described a syndrome or a coincidence”.

In an extensive review of the literature, Chong et al identified 105 patients described as suffering from “Hashimoto’s encephalopathy”.⁴² Of these, 13 had no encephalopathy, and three had no thyroid antibodies measured. Of those patients appropriately described and investigated, and in whom an infectious cause had been excluded, 85 patients were identified of whom the majority were female. All had presented with subacute encephalopathy and had raised microsomal, peroxidase, or antithyroglobulin thyroid antibodies at very variable titres. Of these 85 patients, 66% had seizures, 38% psychosis, 27% “stroke like episodes”, 78% raised CSF protein, and, in those patients in whom EEG was performed, 98% had abnormalities. MRI abnormalities were found in 49%, but focal cortical changes were rare. There were variable abnormalities of thyroid function, with 35% having subclinical hypothyroidism. Of 45 patients treated with steroids, 44 improved.

Although Chong et al concluded that this constellation of symptoms was not likely to be a chance finding, they also concluded that there was no evidence that antithyroid antibody concentrations had a role in pathogenesis.⁴² Of the several anti-thyroid antibodies implicated, none has a clear antigenic target within the CNS, and there is no relation between antibody type and clinical phenotype.⁴² Antibody titres in reported cases are often only marginally increased, do not necessarily decline in line with clinical improvement, and are not reliably found in the CSF. High rates of other autoimmune diseases are seen in patients with anti-thyroid antibodies, and in any event up to 20% of normal adults have raised anti-thyroid antibodies.⁴³ It is not surprising therefore that some patients with VGKC related limbic encephalitis, for example, also have raised antithyroid antibodies.³⁴

It is possible that raised thyroid antibodies in “Hashimoto’s encephalopathy” are merely markers for other autoimmune disorders; for this reason it has been proposed that the broader terms including “steroid responsive encephalopathy”⁴⁴ or "steroid-responsive encephalpathy with autoimmune thyroiditis (SREAT)"⁴⁵ be adopted in treatment responsive cases to avoid assumptions of pathogenicity which have not been proven.

NOVEL ANTIBODIES IN LIMBIC ENCEPHALITIS

Despite advances in our understanding of the causes of limbic encephalitis, a number of patients presenting with often classical manifestations are not found to have an infection, an underlying tumour, or VGKC antibodies. The current dendrotoxin assay detects some but not all antibodies to VGKCs, and it is therefore possible that some of these “antibody negative” cases have other antibodies directed against other VGKCs. It is likely however that antibodies against other antigenic targets may also cause the syndrome. A recent paper⁴⁵ accompanied by an editorial,²³ described seven patients with non-herpetic limbic encephalitis, all of whom had temporal lobe signal change either on MR or PET imaging. One patient had VGKC antibodies. Of the remaining six patients, all had abnormal CSF findings; five had an underlying malignancy; five had antibodies against the neuropil of the hippocampus or cerebellum; and one had antibodies against an as yet unidentified intraneuronal antigen. All but one patient responded to immunosuppression or tumour resection. While the phenotype of this syndrome and the antigenic target(s) remain to be fully defined, such cases demonstrate the potential diversity of antibody mediated disease that can cause limbic encephalitis, and the importance of investigating for an occult malignancy—particularly in non-VGKC associated cases.

A CLINICAL APPROACH TO THE MANAGEMENT OF SUSPECTED LIMBIC ENCEPHALITIS

In general:

• Patients presenting with limbic encephalitis may be acutely unwell, and immediate management sometimes requires resuscitation and transfer to the intensive care unit.

• In patients with overt seizures, standard treatment should be initiated without delay; with impaired consciousness, urgent EEG may be required to exclude complex partial status which, if confirmed, should be treated appropriately.

• Even without typical features of Wernicke-Korsakoff syndrome, in patients where alcohol excess or malnutrition is considered, it is prudent to give high dose thiamine supplementation.

• Metabolic disturbances should be excluded, and blood tests should include auto-antibodies, inflammatory markers, and antineuronal antibodies. Positive thyroid autoantibodies should probably best be viewed as a marker of autoimmunity rather than proof of a specific encephalopathy (so-called Hashimoto’s), and should not preclude further investigation for an alternative cause.

• Imaging, ideally with MR, should be undertaken specifically to assess medial temporal lobe signal change.

• Provided it is safe to do so, CSF should be examined, with samples sent for oligoclonal bands, viral PCR, and antineuronal antibodies.

• In patients with or without a fever, especially with a fairly acute onset and a CSF pleocytosis, herpes simplex should be considered, and if in doubt treatment with intravenous aciclovir started.

• Investigation for a broad range of other infective causes including syphilis and HHV6 should be considered, particularly in immunocompromised patients.

• As VGKC associated cases may turn out to be relatively common, and early treatment appears to be associated with a better outcome, testing for these antibodies should preferably be performed early in the course of the disease, especially in the presence of SIADH (samples may be sent for analysis to the Neurosciences Group, Weatherall Institute of Molecular Medicine, Oxford University, John Radcliffe Hospital, Oxford OX3 9DS, UK at a cost of ~£30/sample).

• In patients with evidence for neither an infectious cause nor VGKC antibodies, a paraneoplastic aetiology should be considered and detailed investigation for an occult neoplasm pursued.

An overview of investigation findings in some causes of limbic encephalitis is shown in table 3.

In patients in whom none of these investigations results in a definitive diagnosis, there is little evidence to guide further management. If CNS vasculitis remains possible, brain biopsy may be the only means of diagnosis, and immunosuppression is the treatment of choice. Given the possibility that some patients with unclassified limbic encephalitis may have an as yet unidentified immune mediated cause, a trial of immunosuppression may be considered.²³ There is, however, little evidence to inform the type, dose, or length of treatment that should be given, and in such cases, before starting treatment, consideration should be given to storing CSF and blood for future analysis.