Dynamic prognostication and treatment planning for hepatocellular carcinoma: A machine learning-enhanced survival study using multi-centric data
The Innovation Medicine,
Год журнала:
2025,
Номер
unknown, С. 100125 - 100125
Опубликована: Янв. 1, 2025
<p>A
reliable
system
for
dynamic
prognostication
and
management
of
hepatocellular
carcinoma
(HCC)
is
urgently
needed
but
currently
unavailable.
In
our
previous
work,
we
developed
a
machine
learning
algorithm
termed
"Survival
Path"
(raw-SP)
to
enhance
with
longitudinal
survival
data.
However,
the
model
was
limited
intermediate
stage
HCC
patients,
it
faced
risk
overfitting
due
path
proliferation.
this
study,
novel
framework
incorporating
nodal
fusion
techniques
mitigate
overfitting,
expanded
model's
applicability
all
stages
patients.
A
post-fusion
map
(fusion-SP)
containing
14
different
paths
built,
which
demonstrated
superior
or
non-inferior
accuracy
in
prognosis
prediction
patients
compared
raw-SP,
as
well
traditional
staging
systems
within
first
15
months
since
initial
diagnosis
large-scale
derivation,
internal
external
validation
cohorts.
Subgroup
analysis
showed
fusion-SP
performance
other
models
among
BCLC
C
disease
tumor
burden
above
up-to-seven
criteria.
Under
fusion-SP,
distinct
optimal
combination
treatment
strategies
advanced-stage
at
key
nodes
were
uncovered,
where
frameworks
fall
short.
The
could
serve
robust
tool
facilitating
planning
HCC.
Moreover,
streamlined
methodology
holds
potential
be
applied
across
various
types
cancers.</p>
Язык: Английский
The Value of no Evidence of Disease (NED) in Intermediate‐Stage Hepatocellular Carcinoma After TACE: A Real‐World Study
Liver International,
Год журнала:
2025,
Номер
45(5)
Опубликована: Апрель 15, 2025
ABSTRACT
Background
and
Aims
One‐third
of
patients
with
intermediate‐stage
hepatocellular
carcinoma
(HCC)
can
achieve
imaging‐based
no
evidence
disease
(NED)
during
treatment
after
transarterial
chemoembolization
(TACE)
sequential
therapies;
however,
its
temporal
dynamics,
contributing
factors
prognostic
value
remain
unknown.
Methods
The
longitudinal
data
1665
HCC
from
Sun
Yat‐sen
University
Cancer
Center
were
included
as
a
derivation
cohort;
414
three
external
medical
centers
served
the
validation
cohort.
Image‐Only
NED
is
defined
based
on
imaging
exams
while
having
serum
level
alpha‐fetoprotein
(AFP)
above
upper
limit;
Image‐Bio
pertains
to
an
additional
achievement
normal
AFP.
A
semi‐Markov
multistate
model
was
adopted
identify
transitions
between
intermediate
states,
which
unreached,
NED,
recurrence
death.
time‐dependent
Cox
proportional
hazards
for
overall
survival
(OS)
utilised
evaluate
dynamic
states.
Results
percentage
who
reached
Image
35.2%
24.7%
in
cohort,
37.4%
31.4%
proportion
peaked
by
end
second
year
since
initial
declined
gradually.
Patients
had
higher
risk
compared
subgroup
(
p
<
0.05).
With
unreached
reference,
multivariate
showed
(HR
0.44;
95%
CI
0.33–0.59)
0.26;
0.20–0.33)
significant
states
that
predict
distinct
OS
HCC,
further
confirmed
multi‐centre
Conclusions
Our
study
highlights
clinical
course
demonstrates
significance
TACE.
recommended
serve
important
endpoint
management
HCC.
Язык: Английский