The Application of MicroRNAs in Cancer Diagnostics - RNA Interference from Biology to Therapeutics - page 264

expression patterns accurately reflected the developmental lineage and differentia-
tion stage of tumors and, furthermore, showed that a miRNA expression signature
was superior to an equivalent mRNA signature for determining the tissue of origin
of undifferentiated tumors. Likewise, in an independent report, Rosenfeld et al . used
miRNA expression profiles from microarray-based analysis of 401 clinical samples,
representing 22 different tumor tissues and metastases, to construct a 48-miRNA
molecular classifier that identified the tissue of origin of metastatic tumors of
unknown primary site with high accuracy [ 89 ] . This miRNA-based classi fi er was
designed as a binary decision tree and was reported as superior to previously pub-
lished mRNA-based classifiers designed for the same purpose. A correct and precise
diagnosis is essential for optimal clinical management of such cancers.
In another early miRNA profiling study, Volinia et al . used a custom-made
microarray to measure the expression of 228 miRNAs in 540 cancer and nonmalig-
nant samples from lung, breast, stomach, prostate, colon, and pancreas [ 26 ] . miRNA
expression profiles clearly distinguished tumors from nonmalignant samples, and
the authors identified a general solid cancer signature consisting of 21 miRNAs.
Interestingly, the bioinformatically predicted targets for these miRNAs were statis-
tically significantly enriched for known protein-coding cancer genes, supporting the
hypothesis that key oncogenes and tumor suppressor genes are regulated by aber-
rant miRNA expression in human malignancies. In further support of this, the
authors experimentally validated three target mRNAs in vitro and demonstrated a
significant inverse correlation in miRNA abundance and target protein expression
for one selected miRNA:target pair (upregulated miR-106a and downregulated
retinoblastoma protein). They also identified six distinct cancer-specific miRNA
signatures, one for each of the malignancies investigated.

13.3.2 Examples of Diagnostic/Prognostic miRNAs in Solid

and Hematopoietic Tumors

The studies described above were among the first to demonstrate the potential of
miRNA for cancer diagnostics and to highlight miRNA profiling as a fruitful strat-
egy for biomarker discovery. MicroRNA profiles with the potential to discriminate
between normal and cancer tissue have been identified for all malignancies investi-
gated to date, and may potentially be used for early diagnosis. Cancer-specific
molecular changes are often detectable before pathological and cytological abnor-
malities become clearly apparent. Moreover, for several cancer types, it has now
been shown that miRNAs can be used for classification of tumors into clinically
relevant subgroups. For example, some miRNA expression signatures distinguish
histological or cytogenetic subtypes, while other signatures are associated with rou-
tine prognostic parameters such as tumor stage, grade, and metastasis status.
Consistent with this, prognostic miRNA signatures have been published for numer-
ous solid and hematopoietic malignancies. In addition, many studies have identified
miRNAs that modulate responsiveness to particular anticancer drugs in vitro as well
as in patients. Table 13.2 is a representative list of published miRNAs and miRNA