ESTRO 36 Abstract Book

363 / 1082

S348 ESTRO 36 2017
_______________________________________________________________________________________________
Table 1 : Comparison of segmentation - Similarity
measures.
Conclusion
Inter-observer variability in volume delineation can be
reduced with the use of PET modality in radiotherapy
planning, guidelines and teaching. Eye movement metrics
during contours review could be useful for learning
sessions in radiation oncology.
Illustration 1 : Comparison of eye-tracking registrations
during segmentation review. Interobservers variability in
the identification of anatomical targets and critical
structures may explain delineation variability. Greater
awareness of this problem, thanks to eye registration
ameliorated radiotherapists education.
PO-0673 Healing of the bronchial anastomosis and time
between neoadjuvant radiochemotherapy and surgery.
A. Koryllos1, M. Hammer-Helmig 2, W. Engel-Riedel3, D.
Zalepugas1, E. Stoelben1, C. Ludwig4
1 Krankenhaus Köln-Merheim, Thoracic surgery, Köln-
Merheim, Germany
2 Krankenhaus Köln-Merheim, Radiotherapy, Köln-
Merheim, Germany
3 Krankenhaus Köln-Merheim, Thoracic oncology, Köln-
Merheim, Germany
4 Florence Nightingale Hospital, Thoracic surgery,
Düsseldorf, Germany
Purpose or Objective
Objective of our retrospective study was to evaluate the
healing of the bronchial anastomosis in correlation to the
time interval between neoadjuvant high dose
chemoradiation (>60Gy) and surgery for non-small-cell-
lung cancer patients.
Material and Methods
We investigated 485 (2006-2014) patients with NSCLC and
bronchus-sleeve-lobectomy in our clinic. n:81 patients had
neoadjuvant chemoradiation prior to surgery. n:38
patients had only neoadjuvant chemotherapy and n:366
had noch neoadjuvant therapy. Every bronchial
anastomosis was assessed bronchoscopically on the 7th
postoperative day. The anastomosis healing was divided in
5 different grades:
Grade I: good healing, without necrosis or fibrin
Grade II: fokal mucosal necrosis, anastomosis is stable
Grade III: circular mucosal necrosis and / or ischemia in
the depthof the distal bronchus
Grade IV: transmural bronchial necrosis with instability
Grade V: perforation of the anastomosis, insufficiency
Results
The patients with neoadjuvant chemoradiation had in
13,5% a critical anastomosis (Grade IV-V) in comparison to
5.3% for the neadjuvant chemotherapy group and 6.2% for
the non neoadjuvant group.
We investigated the time after the end of chemoradiation
and the rate of critical anastomosis using the students t'
test. The time between the 6th and 8th postradiation-
week was shown to be the optimal time interval for the
bronchus healing of the anastomosis (Grade I-II). Patients
who were operated before or after the above mentioned
time interval had an increased rate of critical bronchus
healing (Grade III-IV-V, p:0.003). The postoperative
complication rate was 16% after the 42th postradiation day
and 34.6% before the 42th postradiation day.
Conclusion
• A neoadjuvant radiochemotherapy prior to a bronchial
sleeve resection impairs the healing of the anastomosis
through a high rate of lokal ischemia.
• On the contrary to the general recommendation, to
perform a lung sleeve resection 4 weeks after
radiochemotherapy, our data favours an optimal interval
of 6 to 8 weeks.
• The neoadjuvant radiochemotherapy raises the
complication rate after a bronchial sleeve resection.
• Application and dosis of radiation therapy as a
neoadjuvant therapy are handled differently.
PO-0674 SABR for lung tumors of 5cm or more: can
knowledge-based planning detect high-risk treatment
plans?
S. Van 't Hof1, M. Dahele1, H. Tekatli1, A. Delaney1, J.
Tol1, B.J. Slotman1, S. Senan1, V. W.F.A.R.1
1VU University Medical Center, radiation oncology,
Amsterdam, The Netherlands
Purpose or Objective
There is limited data available on the use of stereotactic
ablative radiotherapy (SABR) for lung tumors ≥5cm. We
retrospectively assessed high-risk dosimetric features of
treatment plans from patients (pt) with SABR-related
toxicity after treatment of such tumors, and studied if
dose-volume histogram (DVH) predictions of
RapidPlan[Varian Medical Systems], a knowledge based
planning system, could identify sub-optimal plans in pt
with toxicity.
Material and Methods
We retrospectively analyzed outcomes in 54 pt with
primary or recurrent non-small cell lung cancer measuring
≥5 cm, who were treated between 2008-2014 with 5 or 8
fraction SABR using volumetric modulated arc therapy
(VMAT). Of these, 15/54(28%) pt had ≥G3 toxicity, most
commonly radiation pneumonitis (RP, n=9), fatal lung
hemorrhage (LH, n=3) and pleural effusion (n=2). 3/7 pt
with interstitial lung disease developed RP. Treatment-
related death was considered likely (n=3) or possible
(n=8) in 20% of pt. RapidPlan uses a library of different pt
plans to generate a model that can be used to identify
organs at risk (OARs) which are outliers with respect to the
library population. We made a model of all 54 patients and
assessed whether pt experiencing toxicity were outliers. A
new ‘non-toxicity’ model was then generated that
excluded pt with ≥G3 toxicity, and we assessed if this
model could identify any sub-optimal plans in the 15
toxicity pt. This was indicated by at least 1 of the clinical
DVHs of relevant OAR (CL, ipsilateral lung [IL], proximal
bronchial tree [PBT], esophagus or heart) being located
above the DVH prediction range generated by the model.