Sources to variability in circulating human miRNA signatures

Abstract

An increasing number of studies propose circulating microRNAs (miRNAs) as biomarkers for a large number of human diseases including cancer, cardiovascular diseases, neurologic pathologies and others. To further validate miRNA as biomarkers it is indispensable to understand the variability of circulating miRNAs in healthy individuals. We determined the longitudinal miRNomes of 90 serum samples from the Janus Serum Bank in Norway, which have been stored between 23 and 40 y at -25 °Celsius. We profiled 3 serum samples with microarrays for 30 individuals, each. For each individual the samples were collected with a time interval of approximately 5 y. This design allowed insights into inter-individual variability, age dependent miRNA variability and the impact of storage length and pre-processing. A significant proportion of the miRNome was affected by the age of the blood donor and a not negligible, albeit small, part of the miRNome by the storage time. A substantial part of miRNAs was differentially abundant between individuals, independent of the time when samples were collected. Stepwise filtering of the 529 miRNAs that were detected in the serum samples showed 168 miRNAs with differential abundance depending on the time point analyzed, 56 miRNAs differentially abundant between individuals, and 169 miRNAs with an abundance depending on the sampling procedure. While these groups of miRNAs contain generally interesting and biologically important miRNAs, the remaining 135 miRNAs constitute very promising biomarker candidates as they show an overall low variability between healthy individuals, a likewise overall low variability across a longer life span, and a high independence of the sampling process and the storage length.

Keywords: circulating biomarkers; control signatures; miRNA; serum.

Figure 1A.

Box-plots showing the distribution of age of the blood donors (given in years on the Y-axis) and the distribution of the duration of storage times (given in days on the Y-axis).

Figure 1B.

Scatter plot of the age of the blood donors and of the storage times. For each donor, the samples were collected at three comparable time windows of approximately 5 y as indicated by 3 clusters.

Figure 1C.

Distribution of age and storage times for two subsets of samples. The first subset, which is framed by two orange dashed lines includes samples with minimal differences in storage length and maximal variation in the age range (n = 15). The second subset, which is framed by two blue dashed lines, includes samples with minimal age difference and maximal variation in storage times (n = 15).

Figure 2.

Scatter plot of the coefficient variation of the storage length and the age of the blood donors. MiRNAs are generally more affected by the age of the blood donors than by the storage time. Points are colored with respect to the expression intensity of the respective miRNAs.

Figure 3.

Scatter plot of an analysis of variance (ANOVA) for each miRNA with either the storage time or the individual as response variable. The significance values for each miRNA are given as negative decade logarithm.

Figure 4.

Box-plots for miR-328–5p and miR-144–3p showing the variance of both miRNAs between the 30 individuals and the 3 time points. The different individuals (upper part of the figure) and the different time points (lower part of the figure) are referred to by numbers on the X axis. The variation of intensity values is given on the y-axis. The variation of both miRNA was not significant for the storage time but highly significant between the individuals.

Figure 5.

Box-plots for miR-6740–5p and miR-7847–3p showing the variance of both miRNAs between the 30 individuals and the 3 time points. The different individuals (upper part of the figure) and the different time points (lower part of the figure) are referred to by numbers on the X axis. The variation of intensity values is given on the y-axis. The variation of both miRNA was significant for the storage time but not significant between the individuals.

Figure 6.

Box-plots showing the percentage of bases for miRNAs that varied significantly between individuals indicated in blue and miRNAs that varied significantly between storage times indicated in red. The nature of the nucleotides is indicated on the X axis with reference to the basis adenine, cytosine, uracil, and guanine and the percentage of the nucleotides on the Y-axis.