We read with great interest the results that the expression of circulating microRNA (miRNA) miR-122 was substantially higher in acetaminophen-induced acute liver injury (APAP-ALI) patients, compared to healthy controls, using quantitative real-time polymerase chain reaction (PCR) assays with U6 small nuclear RNA (snRNA) as an internal control, as reported by Starkey Lewis et al. 1 However, serum miRNA expression profiles generated from a large number of human samples by our laboratory indicate that circulating U6 snRNA is not a reliable internal normalizer.
The accuracy of circulating miRNA expression analysis critically depends on proper normalization of the data. Endogenous normalizer specific for circulating miRNAs have not yet been well defined. Although some cell/tissue miRNA normalizers, including U6 and miR-16, have been used in circulating miRNA data analyses, recent studies suggest that cell/tissue normalizers may not serve as circulating normalizers. 2 To identify the miRNAs with the most stable expression in human serum, we have evaluated 117 serum miRNA expression profiles from young, aging, and different disease conditions using a real-time PCR array system, based on a global expression mean normalization strategy. 3 Of 332 miRNAs detected in serum, 58 displayed consistent expression across all samples (Fig. 1A). Our criteria to identify ideal circulating miRNA normalizers includes (1) no statistical difference among all groups, (2) the smallest variation across all samples (standard deviation of |−ΔCT| < 1); and (3) relative high expression, closest to the global mean expression. 3 Three miRNAs, miR-374a, miR-374b, and let-7d, met all criteria (Fig. 1B), which well resemble the mean expression value and are stably expressed in all groups. Therefore, they could serve as ideal endogenous normalizers for circulating miRNAs. However, U6 and miR-16 expression was significantly different among the four groups (Fig. 1C), further supporting the previous notion that they are not reliable internal normalizers. Most important, U6 was differentially expressed between healthy young and aging groups (Fig. 1D). Thus, cautious interpretation of the data reported by Starkey Lewis et al. is warranted. Different normalization methods should be further employed to make sure that findings are robust, irrespective of the way of standardization.
We are convinced that these three miRNAs identified here are the best circulating endogenous controls reported thus far, although more validation in different conditions may still be needed. We recommend that suitable endogenous controls should be selected in light of the study design and research conditions, and that the use of two to three endogenous normalizers together resembling the mean expression value may additionally reduce bias and variation.