Introduction

Autoantibodies against the voltage-gated potassium channel (VGKC) complex measured by radioimmunoprecipitation have been reported in a broad spectrum of immunotherapy-responsive neurological illnesses in children and adults. Initially, low levels (<400 pM) were associated with peripheral nerve hyperexcitability syndromes (mainly acquired neuromyotonia), but higher levels were subsequently found in Morvan's syndrome, and in patients with reversible limbic encephalitis.1^–4 In the first case series, 10 patients with limbic encephalitis had VGKC-complex values from 450 to 5128 pM, but some controls were reported to have low positive titres.5 Many subsequent reports have confirmed that VGKC-complex antibodies are associated with limbic encephalitis and a broader range of phenotypes, and in some patients, VGKC-complex antibodies <400 pM did appear to be clinically relevant.6

Latterly it has become clear that VGKC antibodies are not directed against the VGKC itself, but against other cell surface antigens that form part of the VGKC complex. Antigenic targets include leucine-rich glioma-inactivated protein 1 (LGI1),7 typically associated with limbic encephalitis (LE) and a distinctive seizure disorder, faciobrachial dystonic seizures8 and contactin-associated protein 2 (CAPSR-2), more frequently associated with peripheral nerve hyperexcitability syndromes including Morvan's syndrome (neuromyotonia, neuropsychiatric features, sleep disturbance, dysautonomia).9 Antibodies to Contactin2 are much less frequent with no syndrome-specific phenotype having been identified. However, in a proportion of patients with positive VGKC-complex antibodies, a specific antigenic target has not been defined, suggesting that the antibodies may be directed against other antigens within the complex and raising questions about the antibody specificity and relevance.

To investigate the relevance of VGKC-complex antibodies in routine clinical practice, we performed a retrospective analysis of a large case series of individuals from one institution where antibody testing had been requested as part of the diagnostic work-up. We specifically aimed to address the relevance of low-positive versus high-positive antibody levels, and whether discovery of a positive antibody influenced management.

Serology

Serum VGKC-complex antibodies were tested using a radioimmunprecipitation assay.3 Cell-based assays for CASPR2, LGI1 and contactin 2 antibodies were performed using human embryonic kidney cells transfected with cDNAs encoding the relevant proteins.9

Results

An initial audit of 44 VGKC-complex antibody-positive patients demonstrated that 32 had values <400 pM, of whom 90% were not considered to be autoimmune, whereas 9 of the 12 with higher levels were considered to have an immunotherapy-responsive autoimmune disease. To study this more formally we collected all results from 2008 to January 2012 inclusive.

During this time, the serum of 1053 subjects was tested for VGKC-complex antibodies of whom 55 subjects (5.5%) had one or more positive levels (>100 pM). Thirty-two subjects were only ever low positive; and the remaining 23 had high-positive levels at presentation or on subsequent sampling. A summary of clinical presentation, VGKC-complex, LGI1 and CASPR2 antibodies and treatment responses (where available) in the low- and high-positive groups are given in tables 1 and 2, ordered according to the likelihood of an autoimmune condition.

High-positive patients

Seventeen patients had high levels of VGKC-complex antibodies (>400 pM) ranging from 401 to 6846 pM when first tested, and a further six individuals initially had low values which rose to high on subsequent sampling. The clinical features are summarised in table 2.

The mean age was 59 (SD 19.4) with a male:female ratio of 1.3:1, and median time from symptom onset to VGKC-complex measurement of 30 months (range 0.5–206). On the basis of the clinical syndrome and/or response to immunotherapies, 11 (including one individual fulfilling criteria for paraneoplastic disease) were retrospectively classified as having a definite autoimmune condition. Ten individuals had limbic encephalitis, of whom six of eight tested had LGI1 (5) or contactin-2 (1) antibodies. These 10 patients were treated with immunotherapies and showed a good response to treatment. Nine of the remainder were retrospectively assessed as possibly having an autoimmune condition but were not given immunotherapies. Two other individuals had a neurodegenerative condition, and one had hepatic toxicity.

Overall, the suggested autoimmune categories (figure 1) were related to the antibody titre (figure 2A; p=0.013, Fisher's exact test). It was noted that the median time from symptom onset to antibody testing was six months (range 0.5–120) in those patients considered to have a definite VGKC-complex-related disease compared with 28 months (range 0.5–206) in those classified as possible or unlikely (figure 2B; p=0.04, Kruskal–Wallis test).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Summary of all patients tested for voltage-gated potassium channel antibodies, their diagnostic classification and autoimmune probability.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

(A) The number of patients in each of the given categories divided between high and low voltage-gated potassium channel-complex antibodies (p=0.004, χ², analysed on GraphPad Prism). (B) The duration of symptoms at sampling in patients divided between definite, possible and unlikely or undetermined categories. One-way ANOVA, p=0.04.

Discussion

In this unselected retrospective series of adults seen in a tertiary neurology setting, 5.5% of patients showed a positive VGKC-complex antibody result. While the range of conditions in which antibody testing was requested, was inevitably subject to referral bias and may not be entirely representative of the patients seen in a less specialised setting, when we considered retrospectively the likelihood of an autoimmune basis for the disorders,10 high positive values were more common in conditions well described in the context of VGKC-complex antibodies, such as limbic encephalitis, whereas low positive values were more likely to have an alternative, non-autoimmune or unclear diagnosis. The results confirm that high titres of VGKC antibodies are very likely to be clinically relevant, particularly when tested within a few months of onset, whereas values <400 pM should be interpreted with care. Nevertheless, the relatively high number (4/32, 13%) of low-positive cases with tumours suggests that low-positive VGKC-complex antibody levels may be an onconeural marker in some situations.11^–13

Many of the ‘high-positive’ patients had limbic encephalitis or related syndromes (relapsing encephalopathy, seizures), although one subject had an alternative autoantibody (anti-glutamate decarboxylase (GAD)) that was consistent with their neurological presentation of Stiff Person Syndrome. Although the patients with limbic encephalitis were treated with good responses, neurologists had exercised their clinical judgement in not trying immunotherapy for most of the related syndromes, which, therefore, were defined as ‘possible’ autoimmune. By contrast, the ‘low-positive’ patients had many more neurodegenerative diagnoses, atypical presentations or uncertain diagnoses, and only 12% were considered to have a definite autoimmune condition, and only four were given immunotherapy. Importantly, no patient with a VGKC-complex level <400 pM had limbic encephalitis or epilepsy of unknown aetiology, although such patients have been identified elsewhere,6 and low-positive VGKC levels have been associated with classical autoimmune syndromes in a few children.14 ,15

As so many of the low-positive cohorts had unrelated or alternative diagnoses, these data provide strong evidence that considerably less weight should be put on a low- rather than a high-positive antibody level. An exception to this is in the context of peripheral nerve hyperexcitability consistent with neuromyotonia, where VGKC-complex antibodies at low levels are common1 and do not necessarily show binding to LGI1 or CASPR2 (Vincent, M Kiernan, O Watanabe, unpublished data).

Thus, we suggest that low-positive VGKC-complex antibodies may be secondary to other disease pathologies, and it is possible that they are part of a normal repertoire of autoantibodies found in otherwise healthy individuals, which can, under certain circumstances, rise to detectable levels. It is notable that a significant proportion of the low-positive cohorts was diagnosed with a neurodegenerative disorder and in one case with a genetically defined disease, Allgrove syndrome.16 In keeping with recent reports that VGKC levels can occasionally be elevated in cases of Creutzfeld Jacob disease,17 these cases demonstrate that elevated antibodies can be found in degenerative neurological diseases, when their presence is unlikely to be clinically relevant.

The number of subjects (3/32, 9.4%) who were positive for LGI1 or CAPSR2 in the low-positive VGKC cohort was significantly lower than that (6/20, 30%) of the unselected high-positive subjects (χ² 4.5, p=0.034) and that (6/8, 75%) of the limbic encephalitis subjects tested. However, two patients with a definite clinical diagnosis of autoimmune limbic encephalitis were not positive for LGI1 or CASPR2 antibodies, suggesting that there are other antibody targets within the VGKC complex as suggested previously9 and by others.18 Those with high-positive VGKC-complex and LGI1 antibodies all had limbic encephalitis, confirming that LGI1 antibodies are common in this condition.7

A relatively high proportion (three cases, 10%) of patients with low-positive antibodies had tumours (three with breast cancer, one with endometrial carcinoma) and one further subject had a thymoma that was resected before he presented. This is significantly higher than the expected incidence within the general population (0.004% per annum age in the 40–59 age range; 0.014% per annum in the 60–74 age range, http://www.cancerresearchuk.org/cancer-info/cancerstats) and very similar to another recent study.18 While the long duration of neurological symptoms without overt evidence of malignancy in a significant proportion of cases is reassuring, and screening for malignancy may well have been prompted by other factors, the prevalence of underlying malignancy in the low-positive group as a whole is unknown and should be the focus of further study.13

Recently, Klein et al18 reported the clinical presentations of a larger group of VGKC-complex positive patients showing that a small proportion of low-positive VGKC-complex patients are also positive for LGI1 or CASPR2. Although clinical follow-up of these cases was limited, they suggest that these are more likely to be of clinical relevance. Our findings are broadly compatible with their findings; patients with lower antibody titres were less likely to have a specific antigenic target or to have a probable autoimmune condition, except those with LGI1 or CASPR2 antibodies and/or with hyperexcitability syndromes. However, one patient with idiopathic Parkinson's disease and VGKC antibodies at 358 pM was positive for LGI1 and CASPR2 antibodies but was not considered by the treating neurologist to have a definite autoimmune diagnosis and was not treated as such.

It is not clear why high antibody titres are more associated with central nervous system disorders, often without peripheral dysfunction, while low antibody titres cluster more with peripheral hyperexcitability. Certainly very high VGKC/LGI1 antibodies are associated with limbic encephalitis,7 ,9 as shown here, but lower levels of both can be found in peripheral nerve hyperexcitability. Such examples serve to illustrate that antibody results need always to be interpreted in the clinical context, and that our understanding of the pathophysiological underpinnings of many of these syndromes is still incomplete. Nevertheless, the broad rule of thumb that low-positive titres—particularly without a specific antigenic target—are less likely to be relevant is of potential clinical utility.

In conclusion, in this retrospective series, high-positive VGKC antibodies strongly predicted an autoimmune or possibly autoimmune aetiology, suggesting that the cut-off that we use for a high-positive antibody level (>400 pM) has clinical utility, at least for adults. By contrast, the majority of patients with low-positive VGKC-complex antibodies did not have an identifiable antigenic target, or a syndrome thought to be related to the presence of the antibodies; in many patients, a definite alternative diagnosis was reached, and the result rarely altered management. In some cases, however, low-positive VGKC-complex antibodies may reflect a broader autoimmune tendency. The possibility of an occult—perhaps unrelated—neoplasm should always be considered in individuals with a definite or possible autoimmune syndrome. While our data suggest a possible increased risk of an occult tumour in individuals with isolated low-positive antibodies, irrespective of the clinical syndrome, in most patients, these antibodies are not markers of paraneoplasia and further studies are required to demonstrate whether and which occult tumours should be actively sought. These data suggest that, particularly in the absence of a specific antigenic target or a phenotype associated with VGKC-complex antibodies (eg, limbic encephalitis, faciobrachial dystonic seizures, neuromyotonia), low positive antibodies should be interpreted with care. In case of doubt antibody levels should be rechecked after an interval to ensure they are not rising.