26 / 1082
S13
ESTRO 36 2017
_______________________________________________________________________________________________
Figure 1: Exponential fit of radiomics signature score
versus overall survival for different datasets.
Figure 2: Risk group identification based on the radiomics
signature score for each dataset.
Conclusion
The radiomics signature model is able to predict two-year
survival above the chance level in every validation cohort.
Furthermore, the radiomics signature allows the
identification of low, medium and high risk group patients.
These findings indicate that the signature is robust and
transferable across hospitals.
References
[1] Aerts et al. Nat Commun 2014;5:4006.
doi:10.1038/ncomms5006.
[2] Parmar C, et al. Sci Rep 2015;5:11044.
doi:10.1038/srep11044.
OC-0036 Does androgen deprivation therapy result in
lowering the alpha / beta values in prostate cancers?
N.R. Datta
1
, E. Stutz
1
, S. Rogers
1
, S. Bodis
1
1
Kantonsspital Aarau, Radio-Onkologie, Aarau,
Switzerland
Purpose or Objective
Reports indicate that prostate cancers exhibit
fractionation sensitivity with low
α/β
values similar to
those of late-responding normal tissues. This has resulted
in a number of hypofractionated (HRT) randomized clinical
trials being undertaken in prostate cancer. Some of these
have been reported to be isoeffective in terms of identical
biochemical and/or clinical failure (BCF) rates. This allows
estimation of the
α/β
values using the specified RT dose
parameters, thus avoiding any uncertainty in choice of
values for additional variables pertaining to tumour
kinetics, repair and tumour control probability. The
present study has been undertaken to estimate the
α/β
values for prostate cancer from randomized clinical trials
of conventional (CRT) vs. HRT radiotherapy (RT) reporting
similar 5-year BCF. The influence of various tumour and
treatment variables on the
α/β
estimates was also
examined.
Material and Methods
Randomized clinical trials with similar BCF following CRT
or HRT for prostate cancer were collated following a
detailed database search as per the PRISMA guidelines.
The
α/β
value from each trial was derived using the linear-
quadratic (L-Q) expression,
α/β
(Gy) = (d
HRT
D
HRT
– d
CRT
D
C-
RT
)/(D
CRT
– D
HRT
) where, 'd” and 'D” indicate the
dose/fraction and the total dose respectively of the
corresponding HRT and CRT schedules (represented by
subscripts). Prostate cancers being slow growing tumours,
a time factor was not included in these calculations. RT
dose/fraction (1.8-2Gy for CRT; 2.4-3.4Gy for HRT), total
RT dose (73.8-80Gy in CRT; 60-72Gy in HRT), use of
androgen deprivation therapy (ADT) and tumour risk
categories were considered as possible covariates in a
multivariate regression analysis for the estimated
α/β
value.
Results
Seven
randomized trials including 5,916 patients treated
with CRT (n=2,949) or HRT (n=2,967) were
available from
156 citations searched. Three trials recruited patients in
a single risk category (one for each of low, intermediate
or high-risk), 2 trials included both intermediate and high-
risk patients and 2 trials included all risk categories. One
study in low-risk patients did not use ADT while in others,
21%-100% patients were treated with ADT. The
estimated
α/β
values ranged from 1.33-11.1Gy (5.41+4.0) (Table,
Fig). On multivariate analysis, only %patients receiving
ADT predicted the estimated
α/β
values (model R
2
: 0.992;
coefficient = -0.096, p<0.001).