Biology Reference
In-Depth Information
Detection of glycoprotein tumor antigens, such as carcinoembryonic antigen (CEA), secreted
by the epithelium lining of mucinous lesions, has modest sensitivity and specificity of 75%
and 84%, respectively, for mucinous cystic lesions with CEA >192 ng
/
ml. However, many
mucinous lesions with CEA <192 ng
/
ml are missed using this cutoff
[98,104]
. In addition, the
CEA level in cyst fluid is not predictive of malignancy
[105]
. Furthermore, recent analyses
of DNA mutations in cyst fluid have similarly proved to be disappointing. Measurement of
allelic loss amplitude showed a sensitivity of 67% and a specificity of 66% for mucinous cystic
lesions. The presence of KRAS mutations, although highly specific (96%) for mucinous lesions,
showed a low sensitivity of 45%
[106]
.
This persistent diagnostic uncertainty leads to treatment decisions that result in unnec-
essary life-altering surgeries in patients who have cysts with low grade (LG) dysplasia and
who could be managed by observation. Consequently, more accurate diagnostic tools are
needed to improve the treatment of patients with pancreatic cysts. In particular, the ability to
preoperatively predict the grade of dysplasia is poised to have a direct impact on subsequent
management, since IPMN with LG dysplasia can potentially be managed conservatively,
while those with high grade (HG) dysplasia would require surgical intervention due to the
high probability of invasive neoplasia.
A recent study by Wu et al. applied massively parallel DNA sequencing to evaluate the muta-
tion status of 169 genes commonly altered in human cancers in cystic fluid specimens collected
immediately after resection, and found mutations in the GNAS gene at codon 201 in approxi-
mately two thirds of IPMN, but not in serous cystadenoma (SCA) or MCN
[107]
. The combina-
tion of GNAS and KRAS mutations allowed the researchers to distinguish between SCA and
IPMN with 96% sensitivity (95% CI, 0.91 to 0.99) and 100% specificity (97.5% one-sided CI, 0.92
to 1). The presence of a GNAS mutation in cyst fluid could also discern IPMN from MCN, albeit
with a lower accuracy due to the presence of KRAS mutations in approximately 30% of MCN
specimens. Although diagnostic for IPMN, the GNAS
/
KRAS biomarker combination does not
allow for the prediction of the grade of dysplasia or presence of malignancy. Thus, there is still
an urgent need for identifying cyst fluid biomarkers that would be predictive of HG dysplasia
or invasion, thus enabling the stratification of patients for surgery.
The solution to this clinical dilemma may be provided by the application of miRNAs. The
feasibility of using miRNAs in pancreatic biofluids was evaluated by many research groups
investigating candidates previously shown to be associated with development of pancreatic
cancer. For example, expression of miR-155 was interrogated in pancreatic juice samples and
found to be elevated in 60% of IPMN specimens as compared to none of the disease controls
[108]
. In pancreatic cystic fluid, miR-21, miR-221, and miR-17-3p were able to differentiate the
mucinous cysts from those nonmucinous specimens with p<0.01
[109]
. The key candidate,
miR-21, was able to resolve those diagnostic entities with a median specificity of 76% and a
sensitivity of 80%. Recently, Matthaei et al. used cyst fluid specimens collected after surgical
resection to build a miRNA classifier, which allows prediction of grade of dysplasia within the
lining epithelium of an IPMN and identifies cysts that likely need surgical removal
[110]
. They
comprehensively evaluated expression levels of up to 750 miRNAs in parallel studies, involv-
ing microdissected LG and HG FFPE IPMN specimens and cystic fluid specimens represent-
ing these diagnostic entities. They identified a subset of 18 miRNAs that segregated cysts
typically requiring surgery (e.g., HG IPMN) from less malignant lesions that can be conserva-
tively managed (e.g., LG IPMN and SCA). Subsequently, they were able to develop a logistic
Search WWH ::
Custom Search