ESTRO 36 Abstract Book

586 / 1082

S570 ESTRO 36 2017
_______________________________________________________________________________________________
radiotherapy treatment position with immobilisation
devices, and had subsequently received neck radiotherapy
following chemotherapy. CTVs were delineated according
to the principles of modern ISRT guidelines. A
CTV INRT (involved node radiotherapy) was delineated
following coregistration of the radiotherapy position PET-
CT to the planning CT scan. A CTVdiagPET was delineated by
side-by-side assessment of the diagnostic PET-CT; no
additional CTV expansion was made to account for
uncertainties introduced by the absence of optimal pre-
chemotherapy imaging. CTV INRT and CTVdiagPET were
compared using multiple positional metrics. Repeat
coregistrations and delineations were undertaken for 3
patients to determine the effect of intra-observer
variation. Figure 1 shows the variation in CTV when using
the diagnostic and radiotherapy position PET-CT data
respectively during the delineation process.
Figure 1: Comparison of CTV INRT and CTVdiagPET. A)
represents pre-chemotherapy FDG PET-CT acquired in the
radiotherapy treatment-position, B) represents routine
diagnostic pre-chemotherapy PET-CT with arms up, C-F)
planning CT scan with CTVINRT (blue)(contoured using co-
registered pre-chemotherapy radiotherapy treatment-
position PET-CT) and CTVdiagPET (red) (contoured using
side-by-side assessment of diagnostic PET-CT) in the
coronal plane (C), in the sagittal plane (D), in the axial
plane at the inferior extent of the CTVs (E), in the axial
plane at the superior extent of the CTVs (F).
Results
Intra-observer variability was limited, with delineation of
CTV INRThighly reproducible and slightly lower for CTVdiagPET
(mean DICE 0.88 and 0.8 respectively). Superiorly,
CTV diagPET varied by -10 to + 15mmfrom CTVINRT, with a
mean difference of +0.5mm. Inferiorly, CTVdiagPET varied
by -18 to +6mm compared with CTVINRT, with a mean
difference of +3.8mm.
Comparing CTVINRT and CTVdiagPET in the axial plane, the
mean DICE was 0.74. Mean sensitivity index was 0.75
(range 0.56-0.91), showing that on average 75% of the
CTV INRT was encompassed by the CTV diagPET. The average
and maximum ‘mean distance to conformity’ (MDT) under-
coverage was 2.6mm (range 1-4.8) and 7.4mm (range 1.5-
14.3) respectively.
Conclusion
In the absence of treatment-position PET-CT, CTV
expansion cranially and caudally by 10mm and 18mm
respectively, along with generous contouring in the axial
plane, was required to encompass pre-chemotherapy
disease.
EP-1040 Identifying risk factors for L’Hermitte’s
syndrome after chemo-IMRT for head and neck cancer
H. Laidley1, D. Noble2, G. Barnett 2, R. Jena2, N. Burnet2
1Cambridge University Hospitals, School of Clinical
Medicine, Cambridge, United Kingdom
2Cambridge University Hospitals, Oncology, Cambridge,
United Kingdom
Purpose or Objective
Studies suggest that L’Hermitte’s syndrome (LS) after
chemo-radiotherapy for head and neck cancer patients is
related to higher spinal cord doses and younger age. IMRT
plans limit spinal cord dose, but the incidence of LS
remains high. We aimed to identify other risk factors.
Material and Methods
128 patients treated with TomoTherapy™ between 2008
and 2015 prospectively completed a side-effect
questionnaire 3, 6 and 12 months post-treatment. 45
(35.2%) reported typical LS symptoms (consistent with
Grade 1 CTCAE v4.03 myelitis) at least once, and graded
severity of tingling/electric shock sensations down their
spine. Data on age, diabetes, hypertension, concurrent
systemic therapy, and unilateral versus bilateral neck
irradiation (UNI vs BNI) were collected. Radiotherapy
plans were assessed to compare maximum dose, mean
dose, and absolute and partial volumes receiving 10, 20,
30 and 40 Gy to the spinal cord in LS and non-LS patients.
Univariate analyses of baseline parameters against LS
incidence were assessed with Fisher’s exact test and
student’s t-test. Box and whisker plots were used to
inspect dosimetric parameters against LS incidence.
Variables reaching or trending towards significance were
included in a binary logistic regression model.
Results
The only significant variable on univariate analysis was
diabetes (p = 0.032). 13 patients in our cohort were
diabetic (9 on metformin); only 1 developed LS (OR =
0.13). Concurrent weekly cisplatin did not increase LS risk;
23/61 (38%) patients receiving cisplatin developed LS,
compared with 22/67 (33%) who did not (p = 0.58). V40Gy
was the only dose parameter showing a difference
between LS and non-LS patients, so others were excluded
from logistic regression to prevent co-linearity.
The binary logistic regression showed that higher absolute
volume receiving 40 Gy (V40Gy ) was significant (p = 0.037,
OR = 1.55), despite only 29% of patients with LS, and 28%
of patients without LS receiving any dose to the spinal cord
over 40 Gy. There was also a trend for LS patients to be
slightly younger (mean age 56.3 vs 59.4, p = 0.074), and a
protective effect of diabetes was again seen (p = 0.035).
Patients receiving UNI (p = 0.015, OR = 2.82) were more
likely to develop LS; 42% of LS patients received UNI,
compared to 25% of non-LS patients.