ESTRO 36 Abstract Book

594 / 1082

S578 ESTRO 36 2017
_______________________________________________________________________________________________
distant relapse. Eight patients (21.1%) died of tumour
progression.
The TLGATA was predictive of local recurrence (p = 0.04).
ROC curves analysis revealed a cut-off value of 19.6 for
SUVmax, and 13.7 for SUVmean ATA (AUC 0.72, p=0.03 and
AUC 0.72 p=0.03, respectively). The 2-year DFS rate was
significantly lower in patients with a SUVmax >19.6 (p=
0.001) and with a SUVmean ATA >13.7 (p= 0.02).
ROC curves analysis revealed a cut-off value of 19.6 for
SUVmax, 8.6 for SUVmeanATA and 49.1 for TLGATA (AUC 0.8,
p=0.03; AUC 0.9 p=0.007, and AUC 0.8 p= 0.01
respectively). The 2-year OS rate was significantly lower
in patients with a SUVmax >19.6 (p= 0.004), with a
SUVmeanATA >8.6 (p= 0.03) and TLGATA >49.1 (p= 0.004).
Table 1. [18F] FDG-PET paramenters based on multiple
threshold levels.
Parameters
SUVmax SUVmean MTV
(cc) TLG
Adaptive
threshold
algorithm (ATA)
16.37
(4.41 –
34.53)
9.15 (2.8 –
19.71)
17.21
(1.5 –
61.53)
149.89
(5.3 –
877.85)
SUV41%
16.37
(4.41 –
34.53)
10.50
(2.94 –
21.78)
10.30
(1.02 –
58)
130.90
(3 –
850.86)
SUV50%
16.37
(4.41 –
34.53)
11.45
(3.33 –
23.73)
7.79
(0.45 –
47.11)
107.77
(2.10 –
736.33)
Conclusion
Adaptive threshold-based SUVmean, MTV, and TLG and
SUVmax could have a role in predicting local control and
survival in head and neck cancer patients treated with
chemoradiotherapy.
EP-1058 A multicenter study of carbon-io n RT for
locally advanced olfactory neuroblastomas (J-
CROS1402HN)
H. Suefuji1, M. Koto2 , Y. Demizu3, J. Saitoh4, Y.
Shioyama1 , H. Tsuji2, T. Okimoto3, T. Ohno4, K. Nemoto5,
T. Nakano 4, T. Kamada2
1 Ion Beam Therapy Center- SAGA-HIMAT Foundation,
radiation oncology, tosu, Japan
2 National Institute of Radiological Science Hospital-
National Institutes for Quantum and Radiological
Sciences and Technology, radiation oncology, chiba,
Japan
3 Hyogo Ion Beam Medical Center, radiology, tatsuno,
Japan
4 Gunma University Heavy Ion Medical Center, radiology,
maebashi, Japan
5 Yamagata University Faculty of Medicine, radiation
oncology, yamagata, Japan
Purpose or Objective
The combination with surgery and postoperative
radiotherapy is the most common therapy for locally
advanced olfactory neuroblastomas (ONB), but has a high
incidence of local analyzed the ONB
patients treated by carbon-ion radiotherapy (C-ion RT) in
the Japan Carbon Ion Radiotherapy Study Group study. In
this study, we evaluated the efficacy and safety of C-ion
RT for locally advanced ONBs in Japan.
Material and Methods
Patients with T4N0-1M0 ONBs who were treated with C-ion
RT at 4 institutions in Japan between November 2003 and
December 2014 were analyzed retrospectively. A total of
twenty-one patients (16 male and 5 female; median age,
53 years) with locally advanced ONBs were enrolled in this
study.
Results
Main tumor sites included the nasal cavity in 11 patients
and sphenoid sinuses in 10, respectively. Seven patients
had T4a and 14 had T4b tumors. All 21 patients enrolled
in this study did not have cervical node metastases. The
median total dose and number of fractions were 60.8 Gy
(RBE) and 16 fractions, respectively. Four patients
received neo-adjuvant chemotherapy. The median follow-
up period was 39 months (range, 5–111 months). The 3
year overall survival and local control rates were 88.4%
and 83.0%, respectively. Grade 4 late toxicity was
observed in 3 patients . Of the three patients, 2 developed
ipsilateral optic nerve disorder and 1ipsilateral
retinopathy. With respect to these patients whose adverse
events could not be avoided, the GTV was over 34cc and
the tumors were in close proximity to the orbit. Except
eye disorder, grade≧ 4 late toxicities did not occurred.
Conclusion
C-ion RT is an effective treatment modality for locally
advanced ONB.
EP-1059 A [18F] FDG-PET adaptive thresholding
algorithm for delineation of RT volumes of
head&neck cancer.
M. Paolini1, L. Deantonio1, L. Vigna 2, R. Matheoud2, G.
Loi2, G. Sacchetti3, M. Brambilla2, M. Krengli1
1University Hospital Maggiore della Carità,
Radiotherapy, Novara, Italy
2University Hospital Maggiore della Carità, Medical
Physics, Novara, Italy
3University Hospital Maggiore della Carità, Nuclear
Medicine, Novara, Italy
Purpose or Objective
A standardized way of converting PET signals into target
volume is not yet available. The aim of this study was to
evaluate a [18F] FDG-PET adaptive thresholding algorithm
for the delineation of the biological tumour volume for the
radiotherapy (RT) treatment planning of head and neck
cancer patients.
Material and Methods
Thirty-eight patients, who underwent exclusive intensity
modulated RT with simultaneous integrated boost (IMRT-
SIB) for head-and-neck squamous cell carcinoma (3 oral
cavity, 9 nasopharynx, 19 oropharynx, 6 hypopharynx, and
1 larynx cancer) were included in the present study.
Thirty-five/38 patients presented a locally advanced
disease (92.1%), and 30/38 patients (78.9%) received a
concomitant chemoradiotherapy. For all patients, [18F]
FDG-PET/CT was performed in treatment position with the
customized thermoplastic mask. Two radiation oncologists
defined the primary biologic tumour volumes (BTV) using
the adaptive thresholding algorithm implemented on the
iTaRT workstation (Tecnologie Avanzate, Italy). The
algorithm used specific calibration curves that depended
on the lesion-to-background ratio (LB ratio) and on the
amplitude of reconstruction smoothing filter (FWH).
The evaluation of reproducibility of adaptive thresholding
algorithm for volume estimation was determined by the
volume overlap of multiple segmentation of the same
lesion by two radiation oncologists. Each primary tumour
volume was segmented by the adaptive thresholding
algorithm (BTVATA). The target volumes for the primary
tumours previously delineated on the planning computed
tomography (CT) scan using anatomic imaging (CT and
MRI) (gross tumour volume standard GTVST) and a fixed
image intensity threshold method (40% of maximum
intensity) of [18F] FDG-PET standardized uptake value
(GTV40%SUV) were used to perform a volumetric comparison.
Results
The algorithm generated a tumour volume in all but two
patients. The mean values with standard deviation (SD) of
volumes based on the three different methods were
reported in Table 1.
The BTVATA was significantly smaller than the GTVST (17 vs.
21 cc, p= 0.04); the conformity index (CI) was 0.46, and
the similarity coefficient (DICE) was 0.7 (Sensibility 66%,
specificity 85%). BTVATA is a part of the GTVST.