S459
ESTRO 36
_______________________________________________________________________________________________
Material and Methods
42 patients with inoperable, stage II-III NSCLC were
treated with (chemo)radiotherapy as part of the
(NCT01024829) PET-boost trial. Patients received
escalated doses (≥ 72 Gy) in 24 fractions consisting of
either a homogenous boost to the PTV, or an
inhomogeneous boost to FDG avid (≥50% SUV
max
) areas.
Patients whom could not be boosted received 66 Gy in 24
fractions. All patients received an FDG PET/CT scan 3
months post-treatment, which was registered to the
planning CT. The lung contours minus the GTV were
compared between the SUV on FDG PET and the planning
CT dose. The planning CT dose was adjusted to equivalent
doses in 2 Gy fractions, assuming an α/β = 3 Gy, and then
binned per 5 Gy increments. The SUV was averaged over
all patients per dose bin and a dose response sigmoid was
fit (SUV vs Gy ) to determine upper and lower asymptotes,
as well as the EC50. The linear portion of the sigmoid fit
(17.6%-82.4% asymptotes difference) was applied to
individual patients for linear fitting, yielding the
correlation coefficient and the SUV response to an
increase in dose (slope). All values were reported as
median[interquartile range (IQR)]. All fits were considered
significant with an alpha of 0.05.
Results
A positive relationship was found between SUV and post-
treatment dose. A sample patient with post-treatment
grade 1 pneumonitis is shown in figure 1. The sigmoidal fit
(figure 2) over all patients was significant (chi-squared =
0.01) with an EC50 at 39 Gy, and lower and upper
asymptotes at 0.60 SUV and 1.31 SUV, respectively. The
linear portion of the sigmoidal fit (15- 60 Gy) was found to
be significantly (p ≤ 0.05), highly linear in individual
patients with a median correlation coefficient of
0.93[0.79-0.97]. Four patients did not have significant
linear fits (p-values ranging from 0.05 to 0.13). The
median SUV response per increase in dose (slope) was
heterogeneous with a median of 0.0083[0.005-0.018]
SUV/Gy, implying a 55 Gy to 200 Gy increase needed per
SUV increase within the IQR.
Conclusion
Strong linear and sigmoidal relationships were found
between post-treatment SUV and planned dose. These
results suggest that increasing dose leads to a highly linear
increase of lung damage up to a certain threshold (60 Gy),
before adopting an asymptotic relationship. Patient
response in the linear fitting was heterogeneous with a
greater than 3 fold difference found in the IQR. These
findings may aid in post-treatment response assessment
and toxicity modeling in NSCLC patients undergoing
escalated dosing regimens.
PO-0848 Predictors of patient-reported incontinence
after prostate cancer RT: results from a cohort study
C. Cozzarini
1
, N. Bedini
2
, E. Garibaldi
3
, D. Balestrini
4
, P.
Franco
5
, G. Girelli
6
, I. Improta
7
, F. Palorini
8
, V.
Vavassori
9
, T. Rancati
8
, R. Valdagni
2,8
, C. Fiorino
7
1
San Raffaele Scientific Institute, Radiotherapy, Milano,
Italy
2
Fondazione IRCCS Istituto Nazionale dei Tumori,
Radiation Oncology 1, Milano, Italy
3
Istituto di Candiolo- Fondazione del Piemonte per
l'Oncologia IRCCS, Radiotherapy, Candiolo, Italy
4
Ospedale Bellaria, Radiotherapy, Bologna, Italy
5
Ospedale Regionale U.Parini-AUSL Valle d’Aosta,
Radiotherapy, Aosta, Italy
6
Ospedale ASL9, Radiotherapy, Ivrea, Italy
7
San Raffaele Scientific Institute, Medical Physics,
Milano, Italy
8
Fondazione IRCCS Istituto Nazionale dei Tumori,
Prostate Cancer Program, Milano, Italy
9
Cliniche Gavazzeni-Humanitas, Radiotherapy, Bergamo,
Italy
Purpose or Objective
To assess clinical and dose factors affecting the incidence
of patient-reported urinary incontinence (INC) at three
years after radical radiotherapy (RT) for prostate cancer
of a large group of patients enrolled in a prospective,
multi-centric trial in the period 2010-2014.
Material and Methods
Enrolled patients were treated in seven Institutions at
different prescribed doses with conventional (74-80 Gy at
1.8-2 Gy/fr, CONV) or moderately hypo-fractionated RT
(65-75.2 Gy at 2.2-2.7 Gy/fr, HYPO) in 5 fractions/week.
Several clinical factors were collected for each patient:
comorbidities, drugs, hormone therapies, previous
surgeries, smoking, alcohol, age, and body mass index. In
addition, the prescribed 2Gy equivalent dose (EQD2) was
considered by applying an alpha-beta ratio of 0.8, 3 and
5Gy, according to values recently reported in the
literature. INC was evaluated through the International
Consultation on Incontinence Modular Questionnaire short
form (ICIQ) filled in by the patients at start/end of RT and
every 6 months until 5 years of follow up. In the current
analysis, patients with ICIQ available at 30 and/or 36
months were considered (n=298;); the incidence of INC at
3 years was defined as the occurrence of an ICIQ value >12
at least once between 6 and 36 months. Univariable and
backward multivariable logistic analyses were performed
to build a predictive model.
Results
In total, 298 patients had the required minimum follow-
up; patients with baseline ICIQ>12 (n=3) were excluded
restricting the analysis to 295 patients (CONV: 149; HYPO:
146, 86% treated with IMRT). The median number per
patients of completed questionnaires was 5 (range: 2-6):
the incidence of ICIQ>12 was 5.1% (n=15) with a
prevalence at 30/36 months equal to 4.1%. Main predictors
at univariable analysis were age (p=0.01,OR=1.19),
baseline ICIQ>0 (p=0.056, OR=2.9), previous TURP
(p=0.04, OR=3.8) and EQD2 (p=0.003-0.02, OR=1.12-1.17
depending on alpha-beta). EQD2 calculated with alpha-
beta=0.8Gy showed the best performances in terms of
calibration plot and p-value and was included in the multi-
variable analysis. Final results suggested a two-variable
model including EQD2 (p=0.005,OR=1.13; 95%CI:1.04-1.24)
and age (p=0.011,OR=1.19; 95%CI:1.04-1.37); the model
showed good performances in terms of goodness of fit