Nature of cognitive impairment with VGKCC antibodies

Although the number of VGKCC antibody-positive patients grew rapidly with increased awareness of the disease and availability of testing, only very few studies systematically addressed the cognitive profile and reversibility of memory dysfunction. In an early study of three patients, dysfunction of anterograde memory was accompanied by temporally ungraded retrograde amnesia extending more than 20 years.8 A recent study including 18 patients with VGKC antibodies (9 LGI1, 3 Caspr2, 6 negative for LGI1/Caspr2) focused on verbal learning, verbal memory and recognition, figural memory, figural learning including supra-span learning, and figural recognition, collectively analysing the function of the temporal lobes.9 During the acute disease phase, 16 of 18 patients had memory deficits in either verbal or figural memory, or in both. At follow-up after 5–70 months, 11 of 17 patients still had impaired function of at least one memory domain. Another study evaluated cognitive function of 19 patients with VGKC antibodies (9 LGI1, 1 Caspr2, 9 not tested) using standardised neuropsychological assessment.10 Similarly, the cohort had significant impairment of memory, processing speed and executive function at the initial visit 16–377 days after symptom onset, while language and perceptual organisation were not affected. At visit 2 ranging from 3 to 44 months after the first assessment, processing speed and executive function had normalised, while at least a third of patients were left with permanent anterograde amnesia.10 A third study addressing cognitive impairment in VGKC antibody-associated limbic encephalitis examined 15 patients and found impairments in attentional–executive functions and verbal and non-verbal memory 1–23 months after symptom onset, mostly in two or three domains.11 At follow-up after 1–68 months, 50% remained impaired in one or two domains despite improvement in verbal and non-verbal memory. Cognitive morbidity can be the predominant long-term symptom, as precisely described in the moving article by Dr Morrow.12

Response to treatment with immunotherapy

Positive effects of immunotherapy on clinical outcome were already observed with the very first described patients 15 years ago and confirmed in later studies.9 ,13 ,14 However, although more intense immunotherapy can be beneficial, most VGKC antibody-positive patients with limbic encephalitis are left with significant memory impairment, with the poorest memory function in LGI1 antibody-positive patients, affecting verbal and figural memory and suggesting permanent functional damage.9 It is possible that delayed initiation of treatment contributes to a poor outcome, similar to the experience with a related common form of autoimmune encephalitis, anti-N-methyl-d-aspartate (NMDA) receptor encephalitis.15 ,16 With the classification of FBDS (ie, brief, very frequent events that often involve posturing of the hemiface and ipsilateral arm) as an LGI1 antibody-mediated syndrome, analyses suggest that early treatment can prevent the progression of the disease from this pathognomonic seizure form into the manifest limbic encephalitis with permanent hippocampal damage.15

Already the initial studies on VGKC antibody-associated encephalopathy suggested that the hippocampus is the predominant target structure of the disease, which has later been confirmed with imaging studies. Significantly larger volumes of amygdala and hippocampus were observed in the acute phase of the disease, while hippocampal atrophy commonly develops at later follow-ups, particularly in patients with LGI1 antibodies.3 ,9 ,17 ,18 However, systematic studies investigating imaging abnormalities in VGKC antibody-associated limbic encephalitis and their relationship to cognitive impairment are urgently needed. These should be separately performed for the different specific antibodies (LGI1, Caspr2, others) in larger cohorts of well-characterised patients to identify characteristic memory profiles that could facilitate early recognition of the disease and target parameters for treatment control and outcome.

Hippocampal involvement in VGKCC-related cognitive impairment

Given that the hippocampus represents the key structure for memory formation and cognitive function, progressive atrophy can well explain the cognitive impairment seen in these patients. The underlying mechanisms of disease might include that LGI1 can influence synaptic transmission in the hippocampus by modulating the ratio of synaptic α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) versus NMDA receptors19 and by supporting the maturation of NMDA receptors in the hippocampus.20 Moreover, experiments with LGI1 antibody-containing human serum could enhance the excitability at hippocampal synapses in slice experiments,21 which is believed to facilitate the process of memory encryption,22 thus supporting that VGKCC antibodies are directly pathogenic.

Importance of VGKCC titre levels

In the above-mentioned syndrome of VGKC-related limbic encephalitis, there is usually no controversy about the role of the antibodies if high titres are present. In particular, the strong association of FBDS with LGI1 antibodies or the unusual type of seizure semiology including autonomic or pilomotor seizures in LGI1 patients23 suggests that bizarre, isolated phenotypes of movement or cognition can occur that may have led to consideration of a psychosomatic disease in several previous cases. However, there are difficulties in attributing cognitive impairment to VGKC antibodies in some patients.24 Patients with low-positive antibodies more often had neurodegenerative disorders, atypical presentations or a rather heterogeneous clinical spectrum. Nonetheless, in some patients with low-positive levels, there is no doubt about the autoimmune aetiology, and others with high-positive levels did not have the full clinical picture of limbic encephalitis.24 Moreover, we have seen a number of patients with cognitive impairment and low-positive VGKC radioimmunoassay (RIA) measures that were clearly positive for LGI1 in cell-based assays—similar findings are reported in the literature.14 Some authors refer to a cut-off at 400 pmol/L, also because lower levels were seen in up to 5% of older patients in the community.3 However, a large series suggested that 61% of patients with LGI1 or Caspr2 antibodies would have been overlooked when VGKC antibody levels below this cut-off were excluded.25 Given that LGI1/Caspr2 antibodies can be detectable in the absence of positive VGKC RIAs, and that the specific antibodies much better associate with a certain clinical phenotype, we strongly recommend to always test for LGI1 and Caspr2 in cell-based assays. Some authors go even further and suggest that testing of VGKCC antibodies might be dispensable and the RIAs could be replaced by specific testing for LGI1 and Caspr2 (future studies with larger patient cohorts will have to elucidate whether this comes at risk of overlooking treatable autoimmune aetiologies in some patients). Until then, low VGKCC antibody titres should not be neglected but always be interpreted in the clinical context, depending on the type of cognitive impairment and kinetic of progression.

VGKCC antibodies in dementia

Finally, VGKCC antibodies are increasingly detected in patients with the initial diagnosis of dementia, in particular if onset of dementia is subacute and the course progresses rapidly. There are a number of cases of VGKC antibody-associated encephalopathy in which the symptoms were dominated by severe cognitive impairment and included rapidly progressing extrapyramidal signs, hallucinations and MRI changes in diffusion-weighted images, thus fulfilling the diagnostic clinical criteria for Creutzfeldt-Jakob disease (CJD). Based on these studies, VGKC antibody-associated limbic encephalitis has become an important early differential diagnosis for CJD and vice versa.26 ,27 VGKC antibodies were rarely seen in pathologically confirmed CJD and occurred mainly at low levels, suggesting that they are a secondary phenomenon rather than driving the disease.28 ,29 Remarkable was also the case of a patient with rapidly progressive behavioural and cognitive decline suggestive of frontotemporal dementia.30 An unexplained ‘repetitive facial grimacing’ makes it likely that the patient had LGI1 antibodies, and steroid therapy resulted in sustained improvement. Similarly, there is an increasing number of case reports describing ‘reversible dementia’ with predominant short-term memory deficits associated with VGKC antibodies.31 ,32 As in other cases of VGKC antibody-associated encephalopathy, the initial diagnosis of a primary neurodegenerative disorder had to be reclassified and the deficits were responsive to immunotherapy. These data collectively suggest that the threshold for antibody testing should be low in order to not overlook treatable aetiologies.

Psychiatric symptoms in those with VGKCC antibodies

As the clinical phenotype for the limbic encephalitis associated with LGI1 and Caspr2 antibodies has been delineated over the past decade, it has been recognised that psychiatric symptoms of mood disturbance, in particular depression and anxiety, have been a part of the presentation in a substantial minority of cases (16.5% in one large case series of 316 patients with VGKCC antibodies).25 Changes in behaviour or agitation are also seen in a further proportion of cases, and this may be particularly the case in children. In an analysis of 20 children (ages 4–15, mean 9) with autoimmune encephalitis and VGKCC antibodies behaviour change was reported by psychiatric problems were identified in 45%33 which includes a combination of hallucinations, agitation, mood disorders as reported by the treating neurologist. 0/7 had positive LGI1, Caspr2 or contactin antibodies.

Furthermore, when case reviews of all those testing positive for VGKCC antibodies are made, a significant proportion of cases have a predominantly neuropsychiatric presentation, which is particularly characterised by a depressed mood state. In a case series of 67 consecutive patients with VGKC antibodies, 24 had a particularly florid neuropsychiatric presentation, of whom 62% had either anxiety or depression.34 In all cases, this was in the context of other symptoms, usually confusion, memory loss and seizures in 13/24. The patients with a neuropsychiatric presentation were more likely to have a medium or high titre of VGKCC antibodies. Sleep disorders (hypersomnolence, insomnia or dream enactment) were also present in 33%.35

VGKCC antibodies in psychiatric presentations

The prevalence of VGKCC antibodies in purely psychiatric populations has not been fully explored. One study found a prevalence of Caspr2 antibodies in 0.9% of 2533 patients with a variety of neurological and psychiatric diagnoses, as well as healthy controls.36 There are several case reports of patients with predominantly psychiatric presentations with VGKCC antibodies, most of whom respond positively to treatment with immunotherapy (table 1).

In one study of 213 psychiatric inpatients at the Mayo Clinic tested over a 10-year period, 7 had positive VGKCC antibodies.35 However, the reason for antibody testing in this group of psychiatric patients, who represented only 1.7% of all hospitalised patients, was the presence of other neurological symptoms, such as confusion, or movement disorder. Furthermore, the immunotherapy was only offered in the cases where the patients had other features suggestive of encephalitis.35

It therefore remains unknown whether other patients with a wholly psychiatric presentation also have levels of VGKCC antibodies that may be clinically significant, and then whether removal of the antibodies leads to an improvement in their psychiatric symptoms. Of the case studies listed in table 1, two of three of those with an identified antibody (LGI1) were treated with immunotherapy, and both improved. Five out of seven of those without an identified antibody were also treated and four of those improved, three substantially so. Case series to date have been subject to selection bias, in that only patients where the clinician suspects limbic encephalitis have had their VGKCC antibodies tested. To the best of our knowledge, there have been no studies investigating the prevalence of the range of VGKCC antibodies in wholly psychiatric patient populations, using appropriately matched control samples.

Relevance of non-LGI1/Caspr2 VGKCC antibodies

A particular difficulty in deciding the likely clinical relevance of the VGKCC antibodies in an atypical clinical presentation is the uncertainty of the relevance of a positive VGKCC result in the absence of a specific antibody. Low levels of VGKCC antibodies are seen in healthy control populations, ranging between 3% and 6.9%.3 ,36 A recent study reported a lack of response to treatment with immunotherapy in patients with low levels of VGKCC antibodies42 with the suggestion that VGKCC antibodies in the absence of a particular target are likely to be irrelevant.42 The conclusion from this study is that the presence of a VGKCC antibody in itself is not sufficient to indicate a pathogenic process.

However, studies to identify the antigenic target in those with VGKC antibodies have found a proportion of patients with high levels of VGKCC antibodies and an immunotherapy responsive-encephalopathy do not have antibodies against Caspr2 or LGI1. The assumption is that they have an, as yet unidentified, antigenic target.6 This is estimated at 19% of patients with VGKCC antibodies.5

There are also studies to suggest that VGKCC antibodies in children, even at relatively low titres, are associated with an inflammatory neurological presentation.33 Further cases have been described with LGI1 antibodies and a treatment-responsive illness, in the absence of significant levels of VGKCC antibodies.5 The suggestion therefore is that while the VGKCC antibodies are not themselves pathogenic, they may indicate a non-specific biomarker for neuronal inflammation. Supporting evidence for this would come from the coexistence of VGKCC with other antibodies or with other autoimmune disorders.43

Relevance of VGKC in psychiatric disorders

There are supporting areas of research to suggest that the VGKC is mechanistically relevant for psychiatric disorders.

Single nucleotide polymorphisms (SNPs) in the second intron of the KCNH2 gene, which codes for the one component of the VGKC, have been associated with an increased risk of schizophrenia.44 Furthermore, these SNPs are related to lower IQ score, slower cognitive processing speed, decreased hippocampal grey matter volume, altered memory task functional MRI (fMRI) signals, as well as a positive response to olanzapine in patients with schizophrenia.45

Mutations of CNTNAP2 gene that codes for Caspr2 protein have been identified in patients with treatment-refractory seizures, schizophrenia, autism and intellectual disability.46 ,47

Hyponatraemia is a common association in VGKC antibody encephalitis, affecting the majority of cases.3 ,33 Hyponatraemia has also been long associated with psychiatric illness, and schizophrenia in particular, with reports predating the use of antipsychotic drugs48 and estimated to be present in up to 20% patients with chronic schizophrenia.49 In one case series, 18% of deaths of people with schizophrenia under the age of 53 were associated with hyponatraemia.50 The cause of hyponatraemia in people with schizophrenia has historically been assumed to be as a result of a psychogenic polydipsia, with patients drinking excessive water as a result of delusional beliefs. It is an intriguing possibility that there may be an alternative biological underpinning to this phenomenon.

There are therefore reasons to suspect that a proportion of patients with a purely psychiatric phenotype may have VGKCC antibodies that may be causally related to their illness, and that may therefore be treatment responsive to immunotherapy in the same way as other presentations with these antibodies. To date, no distinctive pattern of psychiatric presentation has been identified.