Validation of nanostring microrna analysis in leukaemic blood

Aim To assess the technical performance of the NanoString nCounter platform by profiling microRNA (miRNA) in CML patients. Methods RNA (lOOng) from 75 peripheral blood mononuclear cell samples was analysed on the nCounter miRNA panel ( n = 800 probes) in seven batches. Samples were hybridised 12-19h. We used nSolver, R, GraphPad Prism and Normfinder software for normalisation and analysis. Results Ninety-eight miRNAs were expressed at >100 copies (invariant normalised, linear counts). These exhibited an inter-sample fold change >10, and maximum fold change 12,315. QC values were: positive control linearity R 2 = 0.99-1, binding density = 0.27-1.03 (acceptable 0.05-2.25) and imaging — 0.88-1 (acceptable >0.75). Replicate sample raw miRNA counts exhibited a high Spearman’s rank correlation coefficient — 0.8678, p 0.0001. Raw data batch effects resolved after invariant miRNA or cyclic loess normalisation. Hybridisation time affected neither background nor reaction efficiency. Discussion Sample processing is more complex for miRNA analysis than mRNA analysis on the NanoString nCounter platform. Nevertheless, normalisation abrogated inter-run differences. The applications of mRNA based diagnostics using NanoString have recently expanded to diverse cancers, but ours is a first attempt at establishing utility for miRNAs. We conclude that NanoString performance is sufficiently robust to explore this advanced RNA counting technology for diverse applications in pathology.