Assessing the sensitivity and specificity of myositis-specific and associated autoantibodies: a sub-study from the MyoCite cohort.

Abstract

Myositis-specific and associated autoantibodies are important biomarkers in routine clinical use. We assessed local testing performance for myositis autoantibodies by comparing line immunoassay (LIA) to protein radio-immunoprecipitation and identifying clinical characteristics associated with each myositis autoantibody in the MyoCite cohort. Serum samples from patients within the MyoCite cohort, a well-characterised retro-prospective dataset of adult and juvenile idiopathic inflammatory myopathy (IIM) patients in Lucknow, India (2017-2020), underwent LIA at Sanjay Gandhi Postgraduate Institute of Medical Science (SGPGIMS), Lucknow. Immunoprecipitation of 147 IIM patient serum samples (125 adult-onset, 22 juvenile-onset) was conducted at the University of Bath, with researchers blind to LIA results. LIA performance was assessed against Immunoprecipitation as the reference standard, measuring sensitivity, specificity, and inter-rater agreement. Univariate and multivariate logistic regression determined clinical associations for specific MSA. Immunoprecipitation identified myositis autoantibodies in 56.5% (n = 83) of patient samples, with anti-Jo1 (n = 16; 10.9%) as the most common, followed by anti-MDA5 (n = 14, 9.5%). While LIA showed good agreement for anti-Jo1, anti-PL7 and anti-PL12 (Cohen's κ 0.79, 0.83, and 1, respectively), poor agreement was observed in other subgroups, notably anti-TIF1γ (Cohen's κ 0.21). Strongly positive samples, especially in myositis-specific autoantibodies, correlated more with immunoprecipitation results. Overall, 59 (40.1%) samples exhibited non-congruence on LIA and Immunoprecipitation, and κ values for LIA's for anti-TIF1γ, anti-Ku, anti-PmScl, anti-Mi2, and anti-SAE ranged between 0.21-0.60. While LIA reliably detected anti-Jo1, anti-PL7, anti-PL12, anti-MDA5, and anti-NXP-2, it also displayed false positives and negatives. Its effectiveness in detecting other autoantibodies, such as anti-TIF1γ, was poor.