ESTRO 36 Abstract Book

332 / 1082

S317 ESTRO 36 2017
_______________________________________________________________________________________________
lesion glycolysis (TLG) and 5) SUVpeak. Univariate and
multivariate Cox regression were employed throughout.
Results
In the univariate analysis, SUVpeak in the primary GTV was
associated with local control (HR=1.084; p=0.03). The
regional nodal SUVpeak was associated with distant
control (HR=1.11; p=0.003). The other tumor volume and
SUV parameters were also associated with clinical
endpoints (p<0.05). In the multivariate analysis on OS, the
model included tumor stage (HR =1.41; p=0.02), ECOG
status (HR=2.92; p<0.001), Charlson comorbidity index
(HR=1.37; p=0.005), pack years of smoking (HR=1.015;
p<0.001) and primary GTV volume (HR=1.0074; p=0.02).
The model on DFS included ECOG status (HR=2.13;
p<0.001) and the union of all GTV volume (HR=1.0080;
p<0.001). These models were not improved by including
SUV parameters. However, analyzing the subgroup of HPV
negative OPC and non-OPC (n=77), the model on OS could
be improved by including MTV (HR=1.018; p<0.001) or TLG
(HR=1.0015; p=0.007). The model on DFS could be
improved by including one of the following parameters:
MTV (HR=1.040; p<0.001), TLG (HR=1.0036; p<0.001) or
SUVpeak of regional nodes (HR=1.15; p=0.001).
Conclusion
Accounting for clinical risk factors in the multivariate
modeling of survival after radiochemotherapy of head and
neck cancer, the SUV parameters appear less important
than found in previous studies. The association between
the SUV parameters and survival seems to be stronger for
HPV negative patients.
PO-0615 Can diffusion-weighted MRI predict for
xerostomia and QoL in head and neck patients receiving
RT?
K. Nguyen 1, M. Min1,2,3, L. Holloway1,2,3,4, M. Jameson1,2,
C. Rumley 1,2, A. Fowler1, M. Lee1,3, D. Forstner 1,2,3, R.
Rai1, G. Liney1,2,3,5
1 Liverpool Hospital, Radiation Oncology, Liverpool,
Australia
2 Ingham Institute of Applied Medical Research, Radiation
Oncology, Liverpool, Australia
3 University of New South Wales, Department of
Medicine, Sydney, Australia
4 University of Western Sydney, Medicine, Sydney,
Australia
5 University of Wollongong, Medicine, Wollongong,
Australia
Purpose or Objective
Diffusion-weighted MRI (DWI) has the ability to identify
specific cellular characteristics of tissues. Although DWI
has not been traditionally used in RT planning, there is
increasing interest regarding its potential worth in
predicting RT-related toxicities. Our study aims to
investigate 1) serial changes in apparent diffusion
coefficient (ADC) maps of salivary glands quantified from
DWI and 2) whether DWI can predict for xerostomia and
quality-of-life (QoL) in patients with mucosal primary
head and neck cancer (MPHNC) receiving primary RT.
Material and Methods
19 patients with newly diagnosed MPHNC receiving RT
were included. DWIs were acquired before, during (week
2,3,5,6) and after RT (post-RT week 4,12). Parotid and
submandibular glands were contoured on DWI and
corresponding estimated ADCs extracted (Figure1). 'Dry
mouth” according to Common Terminology Criteria for
Adverse Events (CTCAE) v4.0 and 'saliva” QoL according to
University of Washington QoL (UW-QoL) questionnaire v4.0
were recorded for a follow-up period of 11-23months post-
RT. A Spearman Correlation test was used to evaluate
associations (p≤0.05=significant) between ADCmean of
ipsilateral/contralateral/bilateral parotid and
submandibular glands and CTCAE, as well as UW-QoL
scores.
Results
Serial ADC mean for parotid and submandibular glands, both
ipsilateral and contralateral to primary cancer, showed an
increasing trend from pre-RT to post-RT (Figure2).
On analysis of parotid glands, a statistically significant
positive correlation existed between xerostomia and
bilateral parotid ADCmean on week5 DWI 0.624(p=0.023),
while a statistically significant negative correlation
existed between "saliva” QoL and bilateral parotid ADCmean
on week5 DWI -0.629(p=0.038). Similarly, on analysis of
submandibular glands, a positive correlation existed
between xerostomia and bilateral submandibular ADCmean
on week5 DWI 0.751(p=0.008), while "saliva” QoL and
change in bilateral submandibular ADC mean between pre-
RT and week5 DWI recorded a significant negative
correlation -0.791(p=0.034).
Conclusion
Our study shows an increasing trend in ADC mean on DWI in
salivary glands pre- to post-RT. As ADC is
an inverse measure of cellular density, this suggests there
is a decrease in the cellularity of salivary glands pre- to
post-RT detectable on DWI.
Correlations between ADCmean and xerostomia were
identified at week5 which suggests ADCmean measurements
at this time-point (approximately mid-RT) may be most
sensitive to predict for late xerostomia, rather than at any
other time-point studied. The significant negative
Spearman correlation between QoL following RT and
change in ADCmean suggests a link between worsening QoL
and increased ADCmean (or decreased cellularity) of salivary
glands.
In conclusion, DWI may be a valuable predictive tool for
late toxicities in MPHNC patients receiving RT. DWI has
potential to aid adaptive radiotherapy which spares
salivary glands depending on areas of high cellularity and
should be further explored.