Use of discriminant analysis in assessing pulmonary function worsening in patients with sarcoidosis by a panel of inflammatory biomarkers

To date, no sufficiently sensitive and specific single marker has been found to predict the clinical course of sarcoidosis. We designed a cohort study to investigate whether a panel of biomarkers measured in bronchoalveolar lavage (BAL) and peripheral blood could help predict pulmonary function worsening during the clinical course of sarcoidosis. We analyzed 30 individuals with histologically proven sarcoidosis. At baseline, participants underwent pulmonary function tests (PFTs), fiberoptic bronchoscopy and radiological investigations. BAL and blood cellular profiles were obtained from all individuals and six pro-inflammatory molecules were quantified in BAL and serum. PFTs were performed at follow-up visits over a 2-year period. Using discriminant function analysis, a canonical variable was generated to optimize the accuracy of selected variables in predicting pulmonary function worsening and was validated on a subset of nine consecutive individuals with sarcoidosis. A combination of 6 markers from BAL was able to predict pulmonary function worsening in 96 % of patients [95 % confidence interval (CI) 84.4-99.81]. We validated the generated formula on a group of nine patients with sarcoidosis, obtaining 77.8 % correct classification (95 % CI 45.3-93.7). Our results show that a combinational approach could contribute to identifying individuals likely to experience pulmonary function worsening, thus helping to decide the correct therapeutic strategies.