Background People exposed to beryllium may develop beryllium sensitisation (BeS) and, in some cases, progress to chronic beryllium disease (CBD).

Objectives The objective of this study was to test the ability of proteomic technology to identify patterns of serum protein biomarkers that allow differentiation between BeS and CBD and thus remove the need for invasive bronchoscopic procedures.

Methods Initially, SELDI-TOF methodology and analysis was performed on serum samples from 30 CBD and 31 BeS patients.

Results This ‘starter set’ yielded two distinct biomarker pattern sets with eight candidate proteins. The first set differentiated between BeS and CBD with 83.3% sensitivity and 82.3% specificity, with 10-fold cross-validation of 75% and 79%, respectively. The second set of biomarkers yielded higher sensitivity (90.0%) and higher specificity (90.3%), with 10-fold cross-validation of 71.7% and 82.3%, respectively. Due to its greater sensitivity and specificity, the second set of biomarkers was used as the framework for differentiating between CBD and BeS in a second set of serum samples from 450 patients with BeS and CBD. When this larger set of samples was subjected to the biomarker framework in a blinded fashion, it yielded a sensitivity of 43.53% and a specificity of 38.93%.

Conclusions Due to these low sensitivity and specificity values, we have concluded that, currently, the unique set of SELDI-TOF derived biomarkers does not possess the qualities that would allow it to differentiate between a CBD patient and a BeS patient using serum protein biomarkers. Future refinements in sample collection or proteomic technology may be needed to improve biomarker discovery.