Biology Reference
In-Depth Information
and low grade (LG). 47 These types of tumors had
previously been classi
8406, and m/z 10098 d were identi
ed as human
neutrophil peptide-1, cysteine-rich intestinal
protein-1 (CRIP1), and S100A6, respectively.
These three proteins were then used to validate
an independent set of 118 samples by immuno-
histochemistry. Of the three proteins, CRIP1
showed the strongest correlation with patient
survival. Additionally, a support vector machine
algorithmwas generatedusing the sevenproteins
that resulted in 98% classi
ed into three groups d
G1, G2, and G3 d according to histology, but in
2004 the World Health Organization eliminated
the middle grade group. Although high-grade
and low-grade tumors could be easily differenti-
ated based on histological characteristics, the
formerly classi
ed grade 2 tumors proved
more challenging. To better differentiate this
group, a support vector machine algorithm was
generated from IMS data from 27 LG tumors
and 21 HG tumors that was then applied to
a set of 31 G2 tumors. In the initial training set,
cross-validation accuracies of 97.82% and
96.54% were obtained for the LG and HG
groups, respectively. Classi
cation accuracy.
PEP TIDES AND PROTEIN DIG ESTS
Endogenous or enzymatically produced
peptides can be analyzed directly from tissue
sections. The latter technique has allowed for
the
cation of the G2
samples designated 23 samples as LG and 8 as
HG. Samples were blindly classi
ed by two uro-
pathologists and the results compared to the IMS
classi
n-
embedded biopsies, 48 opening up a vast bank
of tissues for IMS analysis as the majority of clin-
ical samples are preserved and stored in this
manner. This analysis is accomplished by
analysis
of
formalin-
xed,
paraf
cations. Of these, 78.3% of the LG and
87.5% of
the HG were correctly classi
ed.
Follow-up analyses of the
ed LG
samples showed that three of these patients pro-
gressed to have HG tumors within
five misclassi
rst
subjecting tissue sections to deparaf
nization
using xylene and graded alcohols before
applying heat-induced antigen retrieval,
commonly performed in histology laboratories.
Trypsin (or another enzyme) is then applied to
the surface of the tissue by either spraying or
spotting before application of matrix. Peptides
can be analyzed from these microdigestion spots
and serve as sequence tags for the parent
proteins, thus allowing for analysis of higher
molecular weight proteins than those that
are typically accessible by traditional protein
imaging. 49 Additionally in many cases,
peptides d either endogenous or enzymatically
produced d can be sequenced and identi
ve years.
The one misclassi
ed HG sample showed no
tumor recurrence within
five years of follow-up.
IMS has recently been applied to a study of
gastric cancer biopsies for proteins that correlated
with patient outcome. 42 A cohort of 63 samples
were used for a training set, 47 with good prog-
nosis and 16 with poor prognosis, to determine
proteins that could be correlated with outcome.
All samples were subjected to MALDI IMS and
then regions of interest corresponding to tumor
were extracted and a subset of these spectra
used for statistical analysis. Analyses were
carried out using ClinProTools
(Bruker) and
the Statistical Analysis of Microarrays plugin for
Microsoft Excel
ed
directly from tissue sections through tandem
mass spectrometry without further isolation or
puri
. A signature of seven
proteins dm/z 3445, m/z 6278, m/z 8406, m/z
8453, m/z 10098, m/z 11353, and m/z 11613 d was
found that correlated with a nonfavorable effect
on patient survival and could not be de
cation.
We have recently shown the application of the
combination of histology-directed pro
ling and
tissue tryptic digestion in the study of Spitzoid
skin lesions. 50 Spitz nevi (SN) are benign skin
lesions that often occur on the face or lower leg
nitively
attributed to any other patient or tumor charac-
teristics. Three of these proteins dm/z 3445, m/z
Search WWH ::




Custom Search