ESTRO 36 Abstract Book

239 / 1096

S226 ESTRO 36
_______________________________________________________________________________________________
Conclusion
Despite its retrospective nature this analysis shows a
significant impact of CRT dose on OS. This can explain the
conflicting results of randomized trials on adjuvant CRT in
PAC in which doses < 45 Gy were generally used.
OC-0427 Prediction models in rectal cancer: an
update of a pooled analysis of 3770 randomized
patients
V. Valentini1, C. Masciocchi 1, J. Van Soest2, G. Chiloiro1,
E. Meldolesi1, M. Gambacorta1, J. Gerard3, S. Ngan 4, J.
Bosset5, A. Sainato6, A. Damiani1, N. Dinapoli1, P.
Lambin2 , A. Dekker2, C. Roedel 7
1 Università Cattolica del Sacro Cuore -Policlinico A.
Gemelli, Radiation Oncology Department, Rome, Italy
2 Maastricht University Medical Center, Radiation
Oncology MAASTRO-GROW School for Oncology and
Development Biology, Maastricht, The Netherlands
3 Unicancer- Centre Antoine Lacassagne, Radiotherapy,
Nice, France
4 Peter MacCallum Cancer Centre, Division of Radiation
Oncology, Melbourne, Australia
5 Besançon University Hospital J Minjoz, Radiation and
Oncology, Besançon, France
6 Azienda ospedaliera Universitaria Pisana, Radiotherapy,
Pisa, Italy
7 Goethe University Frankfurt, Radiotherapy and
Oncology, Frankfurt am Main, Germany
Purpose or Objective
In the last years, several prognostic and predictive models
(PMs) for locally advanced rectal cancer (LARC) patients
(pts) have been developed. Aim of this study was to
update the previous PMs [1] developed for local
recurrence (LR), distant metastases (DM) and overall
survival (OS) at 2, 3, 5 and 10 years based on a more
copious pooled set of LARC pts.
Material and Methods
The PMs were developed using the data of the following
LARC trials: Accord 12/0405, EORTC 22921, FFCD 9203,
CAO/ARO/AIO-94, CAO-ARO-AIO-04, INTERACT, I-CNR-RT
and TROG 01.04. Pts were selected applying the following
exclusion criteria: neoadjuvant and adjuvant oxaliplatin
based chemotherapy, no surgery procedure, short-course
radiotherapy and no neoadjuvant radiotherapy. As the
current pooled dataset contains different trials, we used
20% of the data (stratified per trial) as a validation
dataset. Due to variable influence over time, a logistic
regression model was used. Follow-up times (2, 3, 5 and
10 years) for the survival outcomes (LR, DM and OS) were
used as the model outcome. Variable selection was
performed using a stepwise Akaike's information criterion
(AIC) feature selection to determine the optimal subset of
covariates and nomograms developed as a visual
representation. The nomogram shows only significative
covariates (p<=0.01). According to the TRIPOD [2], all Pms
were validated using external validation of type 2b. The
models performance was evaluated using the Area under
the Receiver Operating Curve (AUC) and the brier score.
Results
Three thousand seven hundred seventy patients out of
7612 patients in this pooled dataset satisfied the inclusion
criteria and were analyzed in this study. For each outcome
(LR, DM and OS) performance of training and validation
models, in terms of AUC and brier score were shown in
table 1. Nomograms were generated for each outcome
(LR, DM and OS) at 2, 3, 5 and 10 years. Furthermore as
an example we have reported the new distant metastases
nomogram at 5 years obtained (Figure 1).
Conclusion
The logistic regression models performed with AUC values
always higher than 0.7. The AUC higher in validation than
in training would need further investigation. Nomograms
will be totally showed at the conference.
[1] V. Valentini et al;Journal Clinical Oncology; 2011 [2] S.
Gary et al; Research reporting method; 2015
OC-0428 Surgical time to increase pCR in rectal
cancer: pooled set of 3078 patients from 7 randomized
trials
G. Chiloiro1, C. Masciocchi1, J. Van Soest2, E. Meldolesi1,
M. Gambacorta1, J. Bosset3, J. Doyen4, J. Gerard4, S.
Ngan5, C. Roedel6, F. Cellini1, A. Damiani1, N. Dinapoli1,
P. Lambin 2, A. Dekker2, V. Valentini1
1Università Cattolica del Sacro Cuore -Policlinico A.
Gemelli, Radiation Oncology Department, Rome, Italy
2Maastricht University Medical Center, Department of
Radiation Oncology MAASTRO-GROW School for Oncology
and Development Biology, Maastricht, The Netherlands