Surface Roughness of Asbestiform Fibers and Cleavage Fragments of Amphibole Minerals: Characterization and Importance for Toxicological Evaluation

Abstract Various types of elongate mineral particles have different characteristics that can impact their biological effects and should be considered during risk assessment. Recently, a lot of attention has been paid to the morphological habit of elongate mineral particles, including a classification for asbestiform fibers and nonasbestiform cleavage fragments. Differentiating particle characteristics by habit is essential for toxicology, in light of studies showing that asbestiform fibers are biologically active and carcinogenic, unlike nonasbestiform varieties that appear to be noncarcinogenic. In this study, Atomic Force Microscopy (AFM) was utilized to determine the surface roughness of 26 elongate mineral particles, including 15 asbestiform samples and 11 from nonasbestiform types. It was determined that the mean surface roughness value SZ for asbestiform particles was 147.7 nm (StDev 83.2), and for nonasbestiform, was 339.1 nm (StDev 156.9). A Student’s t test for two groups is statistically significant: t = 4.03, p = 0.0005. Correlation analysis revealed that surface roughness has a moderate negative correlation with the fraction of long, thin particles which are longer than 5 μm and not thicker than 0.15 μm (EMPA) (R = 0.51, R2 = 0.26, P < 0.007). This finding is notable because EMPA was previously established as a predictor of mesothelial carcinogenicity. Combining SZ and another coefficient with toxicological implications (Dimensional Coefficient of Carcinogenicity or DCC) provides the effective discrimination of particles as asbestiform or nonasbestiform, yielding a 0% error rate with an upper bound of 14.7%. In vitro study demonstrated that the viability of human mesothelial cells exposed to elongate mineral particles is increasing with their surface roughness and decreasing with iron content. Additionally, the machine learning approach using a Convolutional Neural Network (CNN) successfully distinguished asbestiform fibers from cleavage fragments via AFM image processing. Future studies are needed to elucidate the full toxicological mechanisms by which surface roughness impacts the predicted habit and carcinogenicity of elongate mineral particles.