Joint mixture effects and population attributable burden of ANA/ENA panels in connective tissue disease-associated interstitial lung disease.

Abstract

Antinuclear antibody (ANA)/extractable nuclear antigen (ENA) panels in connective tissue disease-associated interstitial lung disease (CTD-ILD) are usually interpreted antibody-by-antibody, leaving the joint mixture effect, statistically attributable burden, and operational stratification value unquantified. In 4771 ILD patients with a complete 12-target ENA panel at a tertiary respiratory center over 32 months, sequential logistic models with Benjamini-Hochberg adjustment, mixture analysis, dual-antibody mutual adjustment, E-values, and Greenland-Drescher adjusted population attributable fractions (PAF) were applied. Three antibodies were significant in the fully adjusted model at q < 0.05: anti-Ro52 (adjusted odds ratio [OR] 3.21, 95% CI 2.38-4.32), anti-SS-A (2.77, 1.98-3.86), and anti-Scl-70 (2.65, 1.67-4.22). Positivity for any one of these three antibodies captured 68.3% of CTD-ILD cases at a 31.2% cohort flag rate, with specificity 70.5%, negative predictive value 98.1% and positive likelihood ratio 2.31, providing an operational rule for prioritizing multidisciplinary discussion (MDD) referral. A 5-antibody mixture quintile test yielded a Q5 vs. Q1 OR of 2.54 (95% CI 1.68-3.83, P-trend = 6.6 × 10-11). The fully adjusted OR of anti-SS-A attenuated from 2.77 to 1.85 after additional adjustment for anti-Ro52, suggesting partial mediation of the SS-A signal through Ro52. Greenland-Drescher PAFs were 35.5% for anti-Ro52, 17.1% for anti-SS-A and 7.0% for anti-Scl-70. These figures are interpretable only under strong, unverifiable causal assumptions and are best read as a stratification yield. Among ENA-tested ILD patients, a simple any-positive rule comprising anti-Ro52, anti-SS-A, and anti-Scl-70 identifies two-thirds of CTD-ILD cases while flagging fewer than one-third of the cohort for prospective rheumatologic evaluation.