Autonomic “myasthenia”: the case for an autoimmune pathogenesis

Of all neurological diseases, the one that virtually all third year medical students get right on final exams is myasthenia gravis (MG). MG is an autoimmune disorder that causes weakness and fatigue of skeletal muscles due to an antibody-mediated attack directed against AChRs at neuromuscular junctions (1). It is easy to learn and remember the pathogenesis, immunology, and treatment of MG because the pieces of the puzzle fit together nearly perfectly. In this issue of the JCI, Lennon et al. now provide intriguing evidence of another putative autoimmune disease of AChRs — autoimmune autonomic neuropathy (AAN) (2). They have shown that immunization of rabbits with a fragment of ganglionic AChR induces a condition that mimics the human disease. AAN is very rare but can be life threatening. It is manifested by autonomic disturbances affecting the sympathetic and parasympathetic nervous systems: abnormal gastrointestinal motility with stalled traffic in the intestines, impaired contraction and striking dilatation of the bladder, and disturbances of blood pressure, sweating, salivation, pupillary reactions, and sexual function (3). Based on these experiments and other evidence (4), Lennon et al. suggest that: (a) an antibody-mediated attack directed against neuronal AChRs of autonomic ganglia may be implicated in at least some cases of AAN; and (b) the autoimmune response in some of these patients may be triggered by a remote neoplasm that incidentally expresses the autoantigen. These new findings provide interesting insights into the concepts of autoimmune and paraneoplastic diseases of the nervous system, and information that may be useful for practical treatment of AAN.

Five criteria for recognizing antibody-mediated autoimmune disease

The list of candidate autoimmune diseases affecting every level of the nervous system is long and rapidly increasing. However, it is not simple to prove that a given disease is autoimmune in nature, and even more difficult to identify the autoimmune mechanisms and molecular targets of the disease process. I have proposed a set of criteria to evaluate the evidence for antibody-mediated pathogenesis in putative autoimmune diseases (5). This Commentary will analyze AAN in light of these criteria, and the implications of the results reported in this issue.

1. Autoantibodies are present in patients with the disease. This is a key starting point but by itself does not prove that a disease is antibody-mediated. Antibodies that bind to the α3 subunit of ganglionic AChRs have been detected in a substantial fraction (41%) of patients with AAN (4), but significant antibody levels have also been detected in patients with other neurological diseases (6). These findings raise the following two questions.

First, why don’t all patients with AAN have anti-α3 AChR antibodies? One rather unlikely possibility is that these antibodies may merely be markers of autoimmunity, not related to the pathogenesis of AAN, and therefore not present in all affected patients. Another alternative is that the antibodies were present early in the disease process and are now gone, or that they are all bound to the targeted AChRs and thereby removed from circulation. Finally, it is likely that at least some patients with a syndrome phenotypically similar to AAN do not have antibodies that bind to α3 AChRs. In some, the disorder may not be autoimmune in nature, while in others, antibodies that react with an antigen closely related, though not identical, to α3 AChR may exist. In the example of MG, approximately 15% of patients with generalized myasthenic weakness lack detectable anti-muscle AChR antibodies (1). Nevertheless, their serum immunoglobulins bind to AChR-expressing mammalian muscle cultures (7), and induce a reduction in AChRs upon passive transfer to mice (8–10). It has recently been shown that sera from approximately 40% of these patients have antibodies that bind to muscle specific kinase (MuSK), a protein that is closely linked to AChR, and plays a role in the clustering of AChRs during development of neuromuscular junctions (11). Whether anti-MuSK antibodies, or some other as yet unidentified antibodies, can account for the loss of AChRs in these patients is not yet known.

Second, why don’t all patients with anti-α3 AChR antibodies have AAN? The Mayo group initially reported that only 5 of 12 patients with significant α3 AChR antibody titers had AAN (6). The other 7 patients had a variety of presumptively autoimmune syndromes: Isaacs syndrome, Lambert-Eaton syndrome, dementia, or sensory neuropathy. It is easy to explain the lack of AAN features in patients with autoantibodies against α3 AChRs, since autoantibodies are well known to be present in individuals without clinical disease. But why did such a high proportion of α3 AChR-antibody–positive patients have other autoimmune diseases? It is likely in at least some of these cases that the associated neoplasm presented multiple antigens that triggered other autoimmune responses (12). It is even conceivable that an immune response to α3 AChR could result in different clinical manifestations in different individuals.

2. Antibody interacts with the target antigen. The “smoking gun” of immunoglobulin bound to ganglionic AChRs has yet to be directly demonstrated in AAN. Nevertheless, there is persuasive circumstantial evidence to support this concept. Previous work has demonstrated that antibodies to α3 AChR bind to AChRs of ganglionic neurons (13). Furthermore, mice genetically lacking α3 AChR demonstrate the same dysautonomic features as patients with AAN (14). Finally, in the present report, autonomic ganglia from α3 AChR–immunized rabbits had electrophysiological features of postsynaptic failure of ACh transmission (2).

3. Passive transfer of antibody reproduces features of disease. This is arguably the most important piece of evidence required to link the disease to antibody-mediated pathogenic mechanisms (15, 16). Ideally, transfer of IgG from human AAN patients with α3 AChR antibodies to experimental animals should reproduce the clinical and electrophysiological features of the disease (4). This would confirm the specific pathogenic role of the antibody, and would rule out a cell-mediated disease mechanism. One would hope that passive transfer studies are in progress.

4. Immunization with antigen produces a model disease. This is the basis of the report by Lennon et al. (2), in which a model of AAN has been elegantly described. Immunization of rabbits with the extracellular domain of α3 AChR resulted in antibody production in most animals, and clinical disturbances of autonomic function consistent with AAN. The fact that the severity of symptoms corresponded with the level of antibodies is further persuasive evidence for the role of α3 AChR antibodies in the disease process.

5. Reduction of antibody levels ameliorates the disease. Rabbits that failed to develop α3 AChR antibodies did not develop disease, and those with low levels were less severely affected. Elsewhere these authors cite examples of a reduction in the severity of AAN in response to treatments that lower the antibody levels (4).

So far, the available evidence provides persuasive, if not yet definitive, support for antibody-mediated pathogenesis of at least some cases of AAN. Future studies are needed to evaluate the effects of passive transfer of serum from both α3 AChR-antibody–positive patients and -antibody–negative patients to recipient animals. From the clinical perspective, these studies highlight the importance of searching for an occult neoplasm in all patients with AAN, and raise the question of whether immunomodulatory treatment should be tried in all patients with AAN, with or without positive α3 AChR antibody results. Ultimately, knowledge of the antigen should lead to antigen-specific therapeutic strategies, like those being developed for MG (17).