ABSTRACT
BACKGROUND: Diagnosis is jeopardised when limited biopsy material is available or histological quality compromised. Here we developed and validated a prediction algorithm based on microRNA (miRNA) expression that can assist clinical diagnosis of lung cancer in minimal biopsy material to improve clinical management. METHODS: Discovery utilised Taqman Low Density Arrays (754 miRNAs) in 20 non-small cell lung cancer (NSCLC) tumour/normal pairs. In an independent set of 40 NSCLC patients, 28 miRNA targets were validated using qRT-PCR. A prediction algorithm based on eight miRNA targets was validated blindly in a third independent set of 47 NSCLC patients. The panel was also tested in formalin-fixed paraffin-embedded (FFPE) specimens from 20 NSCLC patients. The genomic methylation status of highly deregulated miRNAs was investigated by pyrosequencing. RESULTS: In the final, frozen validation set the panel had very high sensitivity (97.5%), specificity (96.3%) and ROC-AUC (0.99, P=10(-15)). The panel provided 100% sensitivity and 95% specificity in FFPE tissue (ROC-AUC=0.97 (P=10(-6))). DNA methylation abnormalities contribute little to the deregulation of the miRNAs tested. CONCLUSION: The developed prediction algorithm is a valuable potential biomarker for assisting lung cancer diagnosis in minimal biopsy material. A prospective validation is required to measure the enhancement of diagnostic accuracy of our current clinical practice.
Subject(s)
Carcinoma, Non-Small-Cell Lung/diagnosis , Carcinoma, Non-Small-Cell Lung/genetics , Lung Neoplasms/diagnosis , Lung Neoplasms/genetics , MicroRNAs/genetics , Aged , Algorithms , Biomarkers, Tumor/genetics , Biopsy , Carcinoma, Non-Small-Cell Lung/pathology , DNA Methylation , Female , Gene Expression , Humans , Lung Neoplasms/pathology , Male , Models, Biological , Models, Statistical , Paraffin EmbeddingABSTRACT
BACKGROUND: Three lung cancer (LC) models have recently been constructed to predict an individual's absolute risk of LC within a defined period. Given their potential application in prevention strategies, a comparison of their accuracy in an independent population is important. METHODS: We used data for 3197 patients with LC and 1703 cancer-free controls recruited to an ongoing case-control study at the Harvard School of Public Health and Massachusetts General Hospital. We estimated the 5-year LC risk for each risk model and compared the discriminatory power, accuracy, and clinical utility of these models. RESULTS: Overall, the Liverpool Lung Project (LLP) and Spitz models had comparable discriminatory power (0.69), whereas the Bach model had significantly lower power (0.66; P=0.02). Positive predictive values were highest with the Spitz models, whereas negative predictive values were highest with the LLP model. The Spitz and Bach models had lower sensitivity but better specificity than did the LLP model. CONCLUSION: We observed modest differences in discriminatory power among the three LC risk models, but discriminatory powers were moderate at best, highlighting the difficulty in developing effective risk models.
