A number of confounding factors can be identified from the search for autoimmune mechanisms over the last 2 decades that may be relevant for future studies. (1) An apparently homogeneous clinical disorder may represent more than one disease process and thereby imply antibody/antigen heterogeneity as, for example, in MG with and without detectable anti-AChR antibodies. In some cases, physiologic studies allow the different forms of the disease to be distinguished as in AIDP and acute inflammatory axonal poly neuropathy. (2) A homogeneous disorder (e.g., LEMS) may have at least two different triggering mechanisms (SCLC and an unknown stimulus). (3) Antigen density may be too low to be detected by the immunohistologic techniques available, as initially occurred in MG and LEMS. (4) Autoantibodies may be detected that are irrelevant to the primary disease, such as anti-striated muscle antibodies in MG. (5) Poor antibody cross-reactivity between species may mean that the pathogenic antibody is undetected in binding assays or in experimental passive transfer studies. For example, anti-AChR antibody in MG shows less than 5% reactivity with Torpedo AChR. (6) A poor regenerative capacity of the target antigen may mean that reduction of circulating autoantibodies by either plasma exchange or ISD treatment is not associated with detectable clinical improvement, as may be the case in SSN in which DRG cells appear to be the target.

Table 5 summarizes the extent to which the data reviewed have established a role for pathogenic antibodies in the light of the postulates for autoimmunity set out earlier and ranks the disorders accordingly. Only in MG with detectable anti-AChR antibody are all the postulates met, including definition of the antigen, experimental passive transfer by the IgG fraction of MG sera, active immunization of experimental animals, and propagation. In both LEMS and the IgMK anti-MAG demyelinating neuropathy the antigen is known, although better characterized in LEMS; the epitopes are not yet denned in either. Data relating to passive transfer are more extensive in LEMS, however; systemic passive transfer of anti-MAG has not yet been reported. In neither condition is an animal model available. In the demyelinating neuropathies, the case for autoimmunity is less complete. Neither in AIDP nor in CIDP is the antigen known, and thus the relevance of the different EAN disorders is uncertain. Current evidence thus rests on the demonstration of serum IgM antibodies that react with peripheral nerve myelin and fix complement and on the intraneural passive transfer studies. In some cases of MG without detectable anti-AChR antibody, the passive transfer data are strongly suggestive of autoantibodies that are directed against determinants of the neuromuscular junction other than the AChR, and that may themselves be heterogeneous. Finally, in SSN there is currently no experimental evidence that the antineural nuclear antibody is directly implicated in the pathogenesis of DRG cell destruction.