Autoantibodies to GPI are predominantly present in extra-articular complications of human rheumatoid arthritis

Several years ago, Benoist et al. described in Cell a T-cell transgenic mouse that spontaneously developed severe arthritis (KRN model). Subsequent reports demonstrated that in this model, disease is caused by antibodies against glucose-6-phosphate isomerase (GPI).

A link between this mouse model and human disease was made when Schaller et al. reported that antibodies against GPI are present in 64% of human rheumatoid arthritis (RA) (Nat Immunol 2001). However, this finding remains controversial, since several other groups could not reproduce these results. Given these apparently conflicting findings, we hypothesized that GPI antibodies are present in a specific subset of RA patients and set out to determine at what point autoantibodies to GPI occur in RA. GPI antibodies were detected in only few sera of of uncomplicated RA (2%)and healthy controls (3%). However, in RA patients with disease manifestation outside their joints, GPI antibodies were more common. 18% of RA patients with skin inflammation (rheumatoid nodules) and 45% of RA patients with vascular inflammation had anti-GPI antibodies. Yet, in RA patients with arthritis complicated by decreased numbers of circulating granulocytes (Felty's syndrome) 92% had GPI antibodies. But, since the GPI antigen was not detected on RA granulocytes, we propose that GPI antibodies are not likely to be directly involved in granulocyte destruction.

In summary, we conclude that anti-GPI antibodies that cause disease in the KRN mouse model are common in human RA complicated by inflammation outside the joint, demonstrating the relevance of the mouse model to human disease.