Diagnostic Value of Induced Sputum in Interstitial Lung Disease

Introduction: Although induced sputum (IS) has recently been used in studies of interstitial lung disease (ILD), there havebeen few reports on studies investigating the usefulness of this method in the differential diagnosis of ILD. The aim of ourstudy was to determine the diagnostic value of differential cell counts and CD4+/CD8+ ratio in induced sputum frompatients with sarcoidosis and other ILDs. Material and methods: We enrolled 59 patients in the study (36 with sarcoidosis, 16 with hypersensitivity pneumonitis[HP] and seven with idiopathic pulmonary fibrosis [IPF]). Sputum was induced a minimum of seven days following BAL byinhalation of 5% NaCl solution for five minutes and repeated four times. Differential cell count was determined in Giemsastained cytospins by counting 400 mononuclear cells for specific cell types per slide. The analysis of T-cell subtypes wascarried out by flow cytometry. The potential for differentiating sarcoidosis from the other ILDs was checked by determiningthe predictive value for CD4+/CD8+ ratio or by using the object classification method based on the k-nearest neighbour(k-NN) rule. The variables used in the k-NN rule were the following parameters of IS: cell viability, total cell count,percentages of macrophages, lymphocytes, neutrophils, eosinophils, CD4+ lymphocytes and CD8+ lymphocytes, and theCD4+/CD8+ ratio. Results: Evaluation of IS was possible in 33 patients (15 with sarcoidosis, 11 with HP and seven with IPF). A CD4+/CD8+ratio in IS exceeding 2.6 had a sensitivity of 100%, and a specificity of 72% with the likelihood of establishing the correctdiagnosis while differentiating sarcoidosis from HP and IPF. On the other hand, when we used the k-NN rule, the likelihood ofestablishing the correct diagnosis was 79% (with an estimated classification error of 21%). Conclusion: Using the object classification method based on the k-NN rule in the differential diagnosis of sarcoidosis, HPand IPF on the basis of all the IS parameters is not associated with a higher likelihood of establishing the correct diagnosisthan the analysis of CD4+/CD8+ ratio alone.