Background Pneumoconiosis is still a problem in workers process non-asbestiform asbestos minerals and serpentinite rocks, such as nephrite, antigorite or talc that may contaminate with paragenetic asbestos minerals. An effective screening method is still lacking. The objective of this study was to assess the diagnostic accuracy using the serum and urinary biomarkers for pneumoconiosis in workers exposed to asbestos-contaminated minerals.

Methods Prediction models of pneumoconiosis were constructed from 140 stone workers (48 cases of pneumoconiosis and 118 controls) exposed to asbestos-contaminated minerals. We measured serum soluble mesothelin-related peptide (SMRP), fibulin-3, carcinoembryonic antigen, and urinary 8-Oxo-2’-deoxyguanosine (8-OHdG)/creatinine levels. Using the ILO international classification of radiographs of pneumoconiosis profusion subcategory ≥1/0 as the reference standard, we established a prediction model by machine learning algorithm. We assessed the accuracy by the area under the receiver operating characteristic curve (AUROC).

Results The SMRP level increased in workers exposed to nephrite. A dose-response relationship was found between the SMRP level and the severity of pneumoconiosis in workers exposed to asbestos-contaminated minerals. Machine learning algorithm composed of sex, age, and 4 serum and urinary biomarkers is able to predict pneumoconiosis with high accuracy (AUROC ranged from 0.76 to 1.00).

Conclusion Our finding highlight the use of serum and urinary biomarkers can be developed as a screening tool for pneumoconiosis in workers exposed to potential asbestos contaminated minerals.