Aim To compare the predictive values of enzyme-linked immunosorbent
assays (ELISAs) and indirect immunofluorescence (IIF) techniques for the detection of
M2 anti-mitochondrial antibodies. Methods Commercial ELISAs are widely
available for the detection of antimitochondrial antibody subtypes in primary biliary
cirrhosis (PBC). We compared the results from two ELISAs (one recombinant, one
purified antigen) with those from two IIF methods in a well-defined cohort of PBC
patients and in patients with systemic lupus erythematosus, Sjögren's syndrome, sicca
syndrome, systemic sclerosis, rheumatoid arthritis and blood donor controls.
Results There was good correlation between a rodent substrate IIF and
ELISA A (r = 0·9134), but poor correlation with ELISA B
(r = 0·5999), which produced many false-positive results in the
control population. We show that rodent IIF alone or human epithelial cell
(HEp-2000®) screening with confirmation by ELISA produce similar
predictive values for PBC and require lesser degrees of skilled interpretation of IIF
patterns. Conclusions We conclude that the specificities of IIF are
greater than the ELISA methods (99% versus 85-97%), although the ELISAs are slightly
more sensitive in biopsy-proven PBC. Careful in-house validation of all new ELISA
technologies is mandatory for good laboratory practice, but IIF in experienced hands
remains an effective and specific assay.
