Diagnostic models of non-small cell lung cancer based on determination of blood cytokines and their receptors

Abstract

Cytokine biomarkers have been suggested to be a promising tool for detecting lung cancer at the initial stage of its development. However, they have not found wide application in practice due to the low diagnostic sensitivity and specificity. The aim of the work was to develop an original combination of the blood level of proteins that could increase the efficiency of their use in the diagnosis of I and II stages of non-small cell lung cancer (NSCLC).152 patients (93 men and 59 women) with newly diagnosed NSCLC were examined: 91 had adenocarcinoma (AC) and 61 had squamous cell lung cancer (SCLC). As a control, 36 healthy patients and 13 patients with hamartoma were examined. The serum level of CYFRA 21-1, SCC, CXCL5 by the immunoassay procedure, of the C-reactive protein (CRP) by the turbidimetric method, and the fluorescence intensity of CXCR2 (MFI CXCR2) receptors on blood lymphocytes by flow cytometry were evaluated.The diagnostic efficacy of the individually analyzed results of measuring CYFRA 21-1, CXCL5, MFI CXCR2, and CRP in AK patients and the level of SCC, CXCL5, MFI CXCR2, and CRP in SCLC was less than 75 %.Two regression equations were developed using a combination of the values of each 4 markers for the diagnosis of the initial disease phase. The ROC-analysis revealed the optimal values of thresholds. In the range 0.307–0.483, the probability of AC on I and II stages was 97.9 %. In SCLC, the threshold range was 0.321–0.529. The predictive value of a positive result was 96.7 %.The examination groups included 17 patients with AK, 14 patients with SCLC, 9 patients with hamartoma, and 12 healthy people. Checking the model performance on an example sample of patients with AC showed that the diagnostic sensitivity was 76.3 %, and the diagnostic specificity was 82.8 %, in SCLC – 76.3 and 81.5 %, respectively