Abstract
Circulating microRNAs in human serum have increasingly been recognized as stable markers for cancer detection. However, there is still a lack of miRNome wide studies over a long period of time with respect to pathogenic processes. We obtained serum samples from the janus serum bank collected prior and after diagnosis of lung cancer. We analyzed the abundance of 904 miRNAs in serum from eight cancer patients at three time points and from six healthy control individuals. Based on the identified miRNA signatures, hierarchical clustering and a self-organizing map identified three major clusters including one cluster containing most of the of the pre-diagnostic samples, a second cluster with mainly post-diagnostic samples, and a third cluster with mainly control samples. Correlation analyses showed that although the profiles were generally stable over several years, most obvious changes of the miRNA pattern seem to occur at a time close to diagnosis. Our findings support the idea that a developing lung cancer might be detectable years prior to diagnosis through a specific miRNA signature and that this signature changes during tumor development.