Analytical variables influencing the performance of a miRNA based laboratory assay for prediction of relapse in stage I non-small cell lung cancer (NSCLC)
1 Medical Prognosis Institute, Venlighedsvej 1, 2970 Hørsholm, Denmark
2 Department of Thoracic Surgery, Roswell Park Cancer Institute, Buffalo, New York, USA
3 Department of Oncology, Odense University Hospital, Sdr. Boulevard 29, DK-5000 Odense C, Denmark
4 Department of Cancer and Inflammation Research, Institute for Molecular Medicine (IMM), University of Southern Denmark, J. B. Winsloews Vej 25, DK-5000 Odense C, Denmark
5 Institute of Pathology, Aarhus University Hospital, Noerrebrogade 44, Bygning 18, DK-8000 Aarhus C, Denmark
6 Department of Pathology, Roswell Park Cancer Institute, Buffalo, New York, USA
BMC Research Notes 2011, 4:424 doi:10.1186/1756-0500-4-424Published: 19 October 2011
Laboratory assays are needed for early stage non-small lung cancer (NSCLC) that can link molecular and clinical heterogeneity to predict relapse after surgical resection. We technically validated two miRNA assays for prediction of relapse in NSCLC. Total RNA from seventy-five formalin-fixed and paraffin-embedded (FFPE) specimens was extracted, labeled and hybridized to Affymetrix miRNA arrays using different RNA input amounts, ATP-mix dilutions, array lots and RNA extraction- and labeling methods in a total of 166 hybridizations. Two combinations of RNA extraction- and labeling methods (assays I and II) were applied to a cohort of 68 early stage NSCLC patients.
RNA input amount and RNA extraction- and labeling methods affected signal intensity and the number of detected probes and probe sets, and caused large variation, whereas different ATP-mix dilutions and array lots did not. Leave-one-out accuracies for prediction of relapse were 63% and 73% for the two assays. Prognosticator calls ("no recurrence" or "recurrence") were consistent, independent on RNA amount, ATP-mix dilution, array lots and RNA extraction method. The calls were not robust to changes in labeling method.
In this study, we demonstrate that some analytical conditions such as RNA extraction- and labeling methods are important for the variation in assay performance whereas others are not. Thus, careful optimization that address all analytical steps and variables can improve the accuracy of prediction and facilitate the introduction of microRNA arrays in the clinic for prediction of relapse in stage I non-small cell lung cancer (NSCLC).