mirnome of human diseases

nature methods | VOL.8 NO.10 | OCTOBER 2011 | 841 only in physiological but also in pathological processes1. Although most reported miRNA expression profiles have been generated from solid tissues, there is growing evidence that miRNA profiles are readily accessible from body fluids, such as blood2,3. The aim of our multicenter study was to elucidate and compare blood expression profiles of 863 miRNAs for different human diseases to test for disease-specific alterations. The generated blood-based ‘miRNome’ data have been deposited in the Gene Expression Omnibus and updated versions are available at http://genetrail.bioinf.uni-sb.de/wholemirnomeproject/. We applied identical standardized experimental and biostatistical procedures to the 454 analyzed blood samples from individuals with lung cancer, prostate cancer, pancreatic ductal adenocarcinoma, melanoma, ovarian cancer, gastric tumors, Wilms tumor, pancreatic tumors, multiple sclerosis, chronic obstructive pulmonary disease (COPD), sarcoidosis, periodontitis, pancreatitis or acute myocardial infarction and from unaffected individuals (controls). All participating centers had to contribute samples to the control group (Supplementary Table 1). The different control cohorts had a high degree of reproducibility between the centers (Supplementary Fig. 1). The platform we used is a highly specific primer extension– based microarray that shows a very small degree of crosshybridization and can be used to distinguish between members of the let-7 family4. To test for technical variance, we repeated the measurements on four samples (two blood samples and two tissue samples) and found a median correlation of 0.97. The correlation between different samples was significantly lower as shown by two-tailed unpaired Wilcoxon Mann-Whitney test (P < 0.05) (Supplementary Fig. 2). To estimate the biological variance, we analyzed blood samples taken from a healthy individual at three different time points during the day (9 a.m., 12 noon and 3 p.m.), with duplicate measurements at each time. Median correlation between the time points was 0.98 and between duplicates it was 0.99 (Supplementary Fig. 3). On average, we found for each disease 103 deregulated miRNAs (P < 0.05; t-test after Benjamini-Hochberg adjustment). A total of 62 miRNAs (7.18% of all 863) were deregulated in at least six diseases in comparison to controls (Supplementary Table 2), and 24 miRNAs (2.78%) were deregulated in >50% of the 14 analyzed diseases. One miRNA (hsa-miR-320d) was deregulated in 11 diseases and three miRNAs (hsa-miR-423-5p, hsa-miR-146b-3p and hsa-miR-532-3p) were deregulated in nine of the tested toward the blood-borne mirnome of human diseases