Analysis of autoantibody profiles in two asbestiform fiber exposure cohorts

Abstract

An increased risk for Systemic Autoimmune Diseases (SAID) was reported in the population of Libby, Montana, where extensive exposure to asbestiform amphiboles occurred through mining and use of asbestiform fiber-laden vermiculite. High frequencies of antinuclear autoantibodies (ANA) were detected in individuals and mice exposed to Libby Asbestiform Amphiboles (LAA). Among the 6603 individuals who have undergone health screening at the Center for Asbestos Related Diseases (CARD, Libby MT), the frequencies of rheumatoid arthritis, systemic lupus erythematosus, sarcoidosis, and systemic sclerosis are significantly higher than expected prevalence in the United States. While these data support the hypothesis that LAA can trigger autoimmune responses, evidence suggests that chrysotile asbestos does not. Serological testing was therefore performed in subjects exposed to LAA or predominantly chrysotile (New York steamfitters) using multiplexed array technologies. Analyses were performed in order to determine a) autoantibody profiles in each cohort, and b) whether the two populations could be distinguished through predictive modeling. Analysis using perMANOVA testing confirmed a significant difference between autoantibody profiles suggesting differential pathways leading to autoantibody formation. ANA were more frequent in the LAA cohort. Specific autoantibodies more highly expressed with LAA-exposure were to histone, ribosomal P protein, Sm/Ribonucleoproteins, and Jo-1 (histidyl tRNA synthetase). Myositis autoantibodies more highly expressed in the LAA cohort were Jo-1, PM100, NXP2, and Mi2a. Predictive modeling demonstrated that anti-histone antibodies were most predictive for LAA exposure, and anti-Sm was predictive for the steamfitters' exposure. This emphasizes the need to consider fiber types when evaluating risk of SAID with asbestos exposure.

Keywords: Asbestos; amphibole; autoantibodies; chrysotile; libby montana.

Figure 1:

Panel of plots examining the distribution of different antibodies (log(x+1) transformation) between the 2 cohorts. Each plot contains 2 box-plots (corresponding to the 2 cohorts, LAA on the left, Steamfitters right) along with individual jittered points along the x-axis for each of the individual observations.

Figure 2:

Graphical display of the strength of evidence from individual Rank-Sum tests. The negative log (base 10) p-value (unadjusted) is displayed on the y-axis and the antibodies are displayed on the x-axis. The different panels represent the different groupings of the antibodies. The solid horizontal line represents a negative log (base 10) of 0.05, indicating that all antibodies with a point above this line had a raw p-value less than 0.05. The dashed horizontal line represents a negative log (base 10) of 0.01, indicating that all antibodies with a point above this line had a raw p-value less than 0.01 The red points indicate antibodies with an adjusted p-value less than 0.05.

Figure 3:

A parallel coordinate plot (PCP) for the 13 ENA antibodies that distinguish LAA from Steamfitters. The plot displays individual patients from combined modeling (n=323) and validation cohort (n=161) divided into LAA group (thin light-blue lines) and Steamfitters group (thin orange lines). A group average is shown as thick blue line for LAA and thick red line for Steamfitters. The y-axis shows antibody measurements (log(x+1) transformation) scaled to 0–1 range.