Neuromyelitis optica and anti‐aquaporin‐4 antibodies: widening the clinical phenotype

Short abstract

Presence of anti‐aquaporin‐4 antibodies in patients with neuromyelitis optica has wide implications

The phenotypic spectrum of the idiopathic inflammatory demyelinating disorders of the central nervous system (CNS) suggests that neuromyelitis optica (NMO), which includes Devic's disease, is a distinct clinical entity separate from classic or conventional multiple sclerosis.¹ NMO is also phenotypically similar to the optico‐spinal form of multiple sclerosis (OSMS) that occurs commonly in Asian populations with multiple sclerosis.² NMO and OSMS, by definition, involve the optic nerves and spinal cord. They tend to be severe diseases, typically causing complete transverse myelitis associated with longitudinally extensive lesions, as seen on magnetic resonance imaging (MRI), which may evolve to form a syrinx. An MRI of the brain is typically normal. NMO is often associated with a mild cerebrospinal fluid (CSF) pleocytosis, which is often polymorphonuclear in nature. Locally synthesised oligoclonal immunoglobulin G (IgG) bands are typically absent in NMO.¹ Pathologically, NMO is characterised by necrosis, eosinophilic and neutrophilic infiltrates, vascular proliferation, and hyalinisation and complement activation in a perivascular “rosette” pattern.¹,³ These clinical and pathological features are uncommon in patients with multiple sclerosis.³

In 2004, the Mayo Clinic group, by using indirect immunofluorescence, reported a characteristic autoantibody staining pattern of CNS tissues with serum from cases with NMO; IgG was shown to outline CNS microvessels of the pia, subpia and Virchow–Robin spaces and co‐localised with laminin.⁴ They aptly named this autoantibody NMO‐IgG.⁴ The sensitivity and specificity of the NMO‐IgG staining pattern in distinguishing cases of NMO and OSMS from related neurological disorders, including cases of conventional multiple slcerosis, were 73% (95% confidence interval (CI) 60 to 86) and 91% (95% CI 79 to 100) for NMO, and 58% (95% CI 30 to 86) and 100% (95% CI 66 to 100) for OSMS. More recently, they have shown that NMO‐IgG binds selectively to aquaporin (AQP)4,⁵ the predominant CNS water channel.

AQPs are a family of membrane‐inserted water channel proteins providing a pathway for osmotically driven water transport through cell membranes. They have a vital role in the reabsorption of water from the renal tubular fluid.⁶ A failure to insert AQP molecules into renal tubular membranes causes nephrogenic diabetes insipidus.⁶ In the CNS, AQP1 is restricted to the apical domain of the epithelial cells of the choroid plexus. AQP4 is expressed on astrocytic foot processes and ependymocytes. AQP9 is localised in tanycytes (hypothalamic bipolar cells bridging the CSF and the hypothalamic portal capillaries) and astrocytic processes.⁷ Messenger RNA expression of AQP3, AQP5 and AQP8 has also been reported to occur in cultured astrocytes.⁷ AQPs in the CNS have a role in osmoreception, potassium siphoning and CSF formation, and are strongly implicated in the pathogenesis of cerebral oedema.⁷

Involvement of the brain has been considered to be an exclusion criterion for the diagnosis of NMO.⁸ Abnormalities on MRI scans of the brain have, however, been described in patients with NMO⁹,¹⁰,¹¹: they are usually non‐specific, but hypothalamic and periventricular lesions may be more specific for NMO.⁹,¹¹ AQP4 expression is not restricted to the optic nerve and spinal cord: the hypothalamic and periventricular distribution of AQP4 seems to correspond with distribution of lesions as evident on the MRI of patients with NMO‐IgG.¹²

In this issue, Nakashima et al¹³ (see p 1073) report that NMO‐IgG was detected in 14 Japanese patients with multiple sclerosis; 12 (63%) had OSMS and 2 (15%) had conventional multiple sclerosis. Not unexpectedly, NMO‐IgG‐positive patients differed from those with conventional multiple sclerosis: longitudinally extensive spinal cord lesions and persistent visual loss (no light perception) were more common in the NMO‐IgG‐positive patients.¹³ Importantly, the two NMO‐IgG‐positive patients with conventional multiple sclerosis had unusual brain lesions, but in other respects had features suggesting OSMS. This and other data support the supposition that OSMS and NMO are the same disease and widen the phenotype of NMO to include cases with brain involvement.

Should NMO‐IgG/or anti‐AQP4 antibody positivity be included as part of the diagnostic criteria for NMO? Until the NMO‐IgG and anti‐APQ4 antibody assays are validated and made widely available, and these findings are confirmed by independent groups, it would be wise to consider anti‐AQP4‐associated neurological disorders to be an emerging clinical entity. It would also be premature to refine the diagnostic criteria for NMO. As soon as a specific biomarker for a well‐defined disease is described, the clinical phenotype associated with the biomarker widens—that is, the so‐called phenotypic spread. This is already occurring; involvement of the brain, particularly the hypothalamus and periaqueductal grey matter, is already considered to be part of the clinical spectrum of NMO.⁹,¹¹,¹⁴ I suspect that cases of idiopathic relapsing myelitis¹⁵ and chronic relapsing inflammatory optic neuropathy¹⁶ will also have to be included in the wider phenotype.

Distinguishing between NMO‐IgG‐positive patients and those with other inflammatory CNS disorders—for example, multiple sclerosis—will be important, as there will be therapeutic implications. NMO tends to be severe, has a strong tendency to relapse and seems to respond to immunosuppressive strategies. At present, long‐term immunosuppression with a combination of azathioprine and corticosteroids is recommended.¹⁷ Similarly, NMO‐IgG‐positive syndromes may prove to be responsive to therapeutic plasma exchange.¹⁸

What are the wider implications of NMO‐IgG or anti‐AQP4 antibodies? It is tempting to speculate that anti‐AQP4 antibodies are functionally relevant and are responsible for the clinical presentation of NMO. Results of passive transfer animal experiments and in vitro functional AQP4 experiments are eagerly awaited. It will also be important to establish an appropriate animal model. This emerging clinical entity raises the question of whether other, possibly non‐immune‐mediated, reversible AQP4 channelopathies exist. Obvious candidates include pre‐eclampsia, eclampsia, posterior reversible encephalopathy syndrome, cyclosporin‐associated encephalopathy and other causes of cerebral oedema, such as that associated with fulminant hepatic failure.