Page 315 - Read Online
P. 315
Shen et al. Evaluation of microRNAs normalization approaches
been broadly used to identify miRNA biomarkers in PCR arrays, miScript, QIAGEN) and SYBR Green
order to better understand the effects of carcinogenic detection. The normalization for miRNA levels usually
exposure, pathogenesis, and cancer risk, as well uses “housekeeping” transcripts, i.e. a reference-
as for early diagnosis and prognostic prediction. gene-based method. Because no universal references
Currently, over 100 mature miRNAs have been found have been accepted by all researches, a variety of
to be dysregulated in hepatocellular carcinoma (HCC) endogenous or exogenous transcripts have been
tissues or blood. [3,4] Many are also associated with selected as references by different microarray and
various HCC risk factors, such as hepatitis B/C virus qPCR assays for data normalization. TLDA includes
[7]
infection, [5,6] aflatoxin B1 exposure, alcohol drinking, a total of 5 endogenous controls (U6 snRNA, RNU44,
non-alcoholic fatty liver disease or non-alcoholic RNU48, RNU24 and MammU6) and miScript uses
steatohepatitis. [8-10] However, only a small proportion spike in cel-miR-39 and 6 references (SNORD61,
of miRNAs (miR-1, miR-9, miR-16, miR-18a, miR- SNORD68, SNORD72, SNORD95, SNORD96A and
21, miR-92a, miR-101, miR-122, miR-199a, miR-221, RNU6B/RNU6-2) as normalizers, while miRCURY
miR-222/223/224, miR-375, miR-483-5p) [11,12] were recommends 5 most stable miRNAs (hsa-let-7i-
consistently confirmed by different studies for their 5p, hsa-miR-222-3p, hsa-miR-425-5p, hsa-miR-93-
role in hepatocarcinogenesis. [13,14] These discrepant 5p, hsa-miR-152) as endogenous references, rather
results may be attributed to a variety of factors than small RNA species (snoRNA and snRNA). Even
that potentially impact miRNA patterns but differ by using the same microarray, different studies may
studies. These factors include the difference in study artificially select various numbers of references to
design (cross-sectional, retrospective or prospective); normalize their results. One study applied miScript
heterogeneity of cancer patients (tumor types, stages, array to profile expression of 84 miRNAs in hepatitis
progression, treatment, hepatitis B, C or mixed viral B virus-related HCC and controls, but only used 2
etiologies); comparison groups (healthy or hepatitis (SNORD61, RNU6-2) out of 6 snRNAs and spiked in
infection controls or non-tumor tissues); types of cel-miR-39 as the normalizer to standardize miRNAs
biospecimens (fresh, frozen or formalin-fixed, paraffin- expression. [16] Another study examined miRNAs in
embedded tissue, serum, plasma, or exosome); and HCC patients and matched controls by the miRCURY
variations in sample collection, preservation and assay using the median of 50% quantile intensity to
processing. Differences of RNA isolation assays, the normalize data. [12] Seven published studies including
input RNA quantity/quality and detection methods can ours have screened miRNA profiles by TLDA in
also impact miRNA expression levels. either HCC tissue or serum/plasma. Two used four
endogenous controls (U6 snRNA, RNU24, RNU44
Even if a careful study design is used and consistent and RNU48) to normalize target miRNA expression in
implementation is applied to pre-analytical and HCC tissues; [17,18] four studies only used one reference
analytical procedures, different methods and (U6 snRNA [11,19,20] or RNU48); [21] and one study did not
“housekeeping” transcripts used to normalize miRNA indicate the reference. [22] More importantly, whether
expression levels may also bias the results and lead those endogenous normalizers are stable among
to misinterpretation of the biological role of miRNAs tested samples is unknown. [11,17-22]
in tumorigenesis. The purpose of normalization is
to remove as much non-biological variations as Global normalization is another strategy which uses
possible to ensure accurate miRNA results within or either the mean or median of detectable miRNAs
between experiments. [15] Therefore, how to select an in each sample as the calibrator to adjust miRNA
appropriate normalizer to adjust miRNA expression expression profiles; this method is adopted from
profiles are crucial to obtaining comparable results. mRNA microarray data normalization protocols. [23]
This is particularly important for epidemiological It is assumed that the mean or the median level of
studies dealing with large data sets usually covering global or most miRNAs is constant across different
multiple experimental batches. tissues or conditions. [24] Although many studies
have demonstrated the advantages of using global
The most common methods to quantitate miRNA normalization, [23,25] the total number of detectable
levels are quantitative real-time polymerase chain miRNAs are much less than mRNAs, which makes
reaction (qPCR) and hybridization microarrays. The it susceptible to extreme values and may bias
methods apply stem-loop reverse transcription and miRNAs expression patterns. [15,26] In addition, large
TaqMan probes (TaqMan low density arrays, TLDA, epidemiological studies usually require independent
Life Technologies) or locked nucleic acid (LNA) validation for a limited number of miRNAs identified
primers (miRCURY LNA™ miRNA arrays, miRCURY, in a discovery set. Practically, it is also not feasible to
Exiqon) or poly (A)-tailed primers (miScript miRNA use global miRNA profiles to normalize expression of
306 Hepatoma Research ¦ Volume 2 ¦ November 18, 2016