Background/Purpose: Anti-transcription intermediary factor 1-gamma (TIF1-g) autoantibody is disease-specific for myositis and identifies a subset of dermatomyositis patients at risk of cancer. Currently non-quantitative immunoprecipitation (IP) is used to identify anti-TIF1-g.  IP has several limitations and development of a quantitative ELISA would improve anti-TIF1-g detection , especially for cancer screening in myositis. Our aim was to develop and validate a quantitative anti-TIF1-g ELISA.

Methods: The ELISA utilizes recombinant purified full length human TIF1-g (Origene Technologies, Rockville MD) coated on the solid surface of a high-binding ELISA plate (Costar, Corning, NY). Patient serum (dilution ≥1:100) was incubated with TIF1-g coated ELISA plates and a horseradish peroxidase conjugated secondary antibody that binds human IgG detected anti-TIF1-g binding.  3,3′,5,5′-tetramethylbenzidine was the horseradish peroxidase enzyme substrate, and the optical density of the resulting chromagen was measured. Units/ml of anti-TIF1-g were determined using a standard serum sample.  Values below the detection range (< 4 U/ml) were considered negative and were assigned a value of 2.  Assay validation utilized lP results as the gold standard for anti-TIF1-g.  Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), ROC curve and area under the curve (AUC) were evaluated.  Mann-Whitney tests were used to compare levels of anti-TIF1-g. Agreement between ELISA and IP was tested. Test-retest reliability was measured.  Myositis patients with positive and negative anti-TIF1-g by IP and non-myositis patients (scleroderma, lupus and RA) and healthy samples were analyzed.

Results:   We identified 55 myositis patients with anti-TIF1-g by IP and 111 myositis patients without anti-TIF1-g from our connective tissue disease database.  Anti-TIF1-g positivity by ELISA significantly correlated with IP results (p<0.001) with strong agreement between both methods (kappa 0.867). Median (IQR) anti-TIF1-g levels in patients with (+) and (-) anti-TIF1-g IP results was 42 (16-95) and 2 (2-2) units/ml, respectively (p<0.001) (figure 1).  The sensitivity, specificity, PPV, NPV and overall accuracy of the anti-TIF1-g ELISA was 91%, 96%, 93%, 95% and 94%, respectively.  The AUC of a ROC curve was 0.938. Test-retest reliability was strong (Pearson r = 0.913, p<0.001). Inter-assay CV :19.5%

Conclusion: We developed a quantitative ELISA for detecting anti-TIF1-g autoantibodies and validated the assay using serum samples from patients with myositis and other autoimmune disorders. The anti-TIF1-g ELISA is simple, sensitive and highly specific. The availability of a validated, quantitative ELISA should improve the detection of anti-TIF1-g antibodies in myositis and malignancies in these patients.

Figure 1. Anti-TIF1-γ antibody levels by ELISA.