887 / 1096
S871
ESTRO 36
_______________________________________________________________________________________________
‘Radiation-induced urgency syndrome’ consists of
different self-reported bowel symptoms. To combine the
symptoms, factor analyses was used. Dose-response
relationships were estimated fitting the data to the Probit
model. ROC curve analyses were used to identify which
organ at risk is correlated the most with the clinical
outcome.
Results
The maximum likelihood estimates of the dose-response
parameters for Probit model, the three organs at risk and
‘radiation induced urgency syndrome’, the Log Likelihood
(LL) value and the AUC are:
D
50
= 51.3 (48.3-54.6),
γ
50
=
1.19 (0.98-1.42), α/β=0.59 (0.034-1.56), LL = -50.2 and
AUC=0.63 for rectum,
D
50
= 51.6 (48.7-54.9) Gy,
γ
50
= 1.20
(0.98- 1.44),
α/β
=2.02 (0.85-4.73), LL = -51.0 and
AUC=0.66 for the sigmoid and
D
50
= 61.4 (56.4-67.5) Gy,
γ
50
= 0.90 (0.73- 1.08),
α/β
= 10.3 (2.1- 1e+06 Gy), LL = -
51.4
and
AUC=0.60
for
small
intestines.
Conclusion
For the studied organs at risk, the dose to the sigmoid is
the best predictor of ‘radiation-induced urgency
syndrome’ among gyneocological cancer survivors. Dose
planners having the ambition to eliminate the syndrome
may consider to delineate the sigmoid as well as rectum
in order to incorporate the dose-response results.
EP-1612 Estimates of the α/β ratio for prostate using
data from recent hypofractionated RT trials.
S. Gulliford
1
, C. Griffin
2
, A. Tree
3
, J. Murray
4
, U. Oelfke
1
,
I. Syndikus
5
, E. Hall
2
, D. Dearnaley
3
1
The Institute of Cancer Research and The Royal Marsden
NHS Foundation Trust, Joint Department of Physics,
London, United Kingdom
2
The Institute of Cancer Research, Clinical Trials and
Statistics Unit, London, United Kingdom
3
Royal Marsden NHS Foundation Trust, Academic Urology
Unit, London, United Kingdom
4
Guy’s & St Thomas' NHS Foundation Trust, Department
of Clinical Oncology, London, United Kingdom
5
Clatterbridge Cancer Centre, Department of Clinical
Oncology, Wirral, United Kingdom
Purpose or Objective
The α/β ratio for prostate cancer has been widely studied
with growing evidence for a value significantly lower than
the standard value for tumours of 10 Gy. Previous studies
have also indicated that there may be a time factor
whereby tumour repopulation should be considered during
the course of radiotherapy[1]. Recent reporting from 4
separate phase III clinical trials comparing
hypofractionated schedules with standard schedules for
prostate radiotherapy allow for further exploration of the
α/β value.
Material and Methods
The α/β ratio for prostate was derived independently for
each of the CHHiP[2], HYPRO[3], PROFIT [4]and RTOG
0415 studies[5] by comparing the outcomes in the
standard and hypofractionated trial arms. This approach
ensures that differences between the trials such as use of
hormones and outcome metrics are accounted for. It was
assumed that the dose response was linear between trial
arms. In 3 of the trials, the hypofractionated schedule was
compared to 2 Gy per fraction, in the RTOG 0415 study the
standard fractionation was 1.8 Gy per fraction. A grid
search approach was used to minimise the error for EQD2.
Repopulation was included in the model using the term
OTT-Tk where OTT is the overall treatment time and Tk is
the number of days from the start of treatment when
repopulation is assumed to begin. A proliferation rate of
0.31 Gy/day was used [1]. The CHHiP trial had two
hypofractionated arms and these were fitted separately.
Results
Figure 1 is a representative example of the grid search
results to minimise the squared difference in EQD2
corrected for outcome between the trial arms. Varying the
Tk parameter has 3 distinct phases; i) Tk less than the OTT
of the hypofractionated arm, where the α/β ratio varies
little ii) Tk between the OTT of the hypofractionated and
standard arms, where the α/β ratio transitions steeply and
iii) Tk greater than the OTT of the standard arm. This last
case reduces to equating the two fractionation schedules.
The best fit parameter values for α/β ratio and Tk are
shown in Table 1 along with the best fit values for the α/β
ratio when repopulation is not considered. For all trials,
the overall best fit parameters included a value of Tk that
was less than the overall treatment time of the standard
arm, indicating an improvement when compared to a
model which considered the α/β ratio only.
Figure 1
Table
1