Introduction and Objective:

piloturuguayanAI and Robotics

Oct 15, 2013 (3 years and 8 months ago)

62 views

Introduction and Objective
:

The long
-
ter
m surveillance mandated for non
-
muscle
invasive bladder
(NMIBC) patients
is
expensive and resource intensive

and no non
-
invasive biomarker currently exists to identify
patients with disease recurrence
. T
he
objective of this study was

to determine if microRNA
(miR) profiling of urine can identify the
presence

of
transitional cell carcinoma (TCC)

in
the
bladder of

patients undergoing surveillance

and to compare it
s test

performance
characteristics to that of
f
lexible
cystoscopy.

Materials and Methods:

We screened 60 patients, which included
30
non
-
recurrers (patients with a history of TCC

but
no recurrence at cystoscopy)
and 30
recurrers (patients with TCC

identified at cystoscopy
)
using a panel of 16

microR
NAs
(miRs) that were over expressed

in bladder cancer

tissue in
previous studies
. Total RNA was extracted from urine
using the mirVana kit (Ambion,
Austin, TX, USA)
and
preselected Taqman miR

assays
used for
profiling by
real time
quantitative polymerase c
hain reaction (
RT
-
qPCR
).

All

reac
tions were performed in triplicate
and the median included in the final analysis. Data analysis was performed using PASW
v18.0
(Chicago, USA)
and
machine learning approach
es
,

to test the predictive abilit
y of
individual and

a combination of miRs profiled in urine
. Specifically

for machine learning
, a
centroid classifier and centroid feature selection procedure was trained using a three
-
fold
cross validation approach and performance was measured using the area under the recei
ver
operator characteristic curve (AUC).

Results:

Individually miR 34a (p
<0.001
, AUC=
0.80
)
,

miR 205 (p
<0.001
, AUC=
0.77
)
,
miR 16
(p=
0.001
, AUC=
0.75
)
,

miR 200c (p=
0.001
, AUC=
0.75
)
,

miR 106b

(p=
0.007
, AUC=
0.7
)
,

miR 221

(p=
0.006
, AUC=
0.71
) and miR 21 (p=
0.00
1
, AUC=
0.70
) were able to distinguish
the recurrers from the non
-
recurrers. However, the best pre
dictor of the presence of TCC

was
achieved using a combination four
miR
s (miR 34a, miR 205
,

miR 16, and miR 200c)
measured in urine (AUC=0.81)
. This combinatio
n of four miR
s in urine was a better
predictor of presence of cancer
in the bladder
than any individual miR
. The difference in miR
expression profiles between the recurrers and non
-
recurrers was highest for patients with
pres
ence of high volume and high
-
gr
ade disease.

Conclusion:

This study suggests that urinary profiling of microRNAs may allow their use in the
surveillance setting

to detect presence of TCC in the bladder as an alternative to flexible
cystoscopy
. Validation of results in an independent cohort is currently underway.