181 / 1082
S168
ESTRO 36 2017
_______________________________________________________________________________________________
This retrospective cohort study includes consecutive OC
patients staged with PET/CT between October 2010 and
December 2014. PET-defined tumour variables and
texture metrics were obtained using ATLAAS, a machine
learning algorithm for optimised automatic segmentation.
A Cox regression model including age, radiological stage,
treatment and 12 novel texture variables was developed
and a prognostic score stratifying patients into quartiles
was calculated. Primary outcome was OS and a p-value
<0.1 was considered statistically significant.
Results
Three hundred and forty-three consecutive patients
[median age=67 (range=24-83), adenocarcinoma=258]
were included. Median survival was 17 months (95% CI
14.685–19.315). Multivariate analysis demonstrated 7
variables that were significantly and independently
associated with OS; age [HR=1.024 (95% CI 1.010-1.038),
p<0.001], radiological stage [HR=1.492 (1.221-1.823),
p<0.001], treatment [HR=2.855 (2.038–3.998), p<0.001],
standard deviation [HR=0.898 (0.815–0.989), p=0.029],
log(coarseness) [HR=1.774 (0.918–3.43), p=0.088],
dissimilarity [HR=1.136 (1.007–1.281), p=0.038] and zone
percentage [HR=0.938 (0.897–0.980), p=0.005].
A
prognostic score derived from the model equation showed
significant differences in OS between quartiles (X
2
=90.13,
df=3, p<0.001).
Conclusion
This study demonstrates the additional benefit of PET
texture analysis in OC staging, over and above the current
TNM system. Our prognostic model requires further
validation, but highlights the potential of texture analysis
to predict survival in OC.
PV-0324 FDG-PET based pelvic bone marrow dose
predicts for blood cell nadirs in CT-RT for anal cancer
P. Franco
1
, F. Arcadipane
1
, R. Ragona
1
, A. Lesca
2
, E.
Gallio
3
, M. Mistrangelo
4
, P. Cassoni
5
, M. Baccega
2
, P.
Racca
6
, R. Faletti
7
, N. Rondi
8
, M. Morino
4
, U. Ricardi
1
1
University of Turin A.O.U. Citta' della Salute e della
Scienza, Department of Oncology- Radiation Oncology,
Torino, Italy
2
A.O.U. Citta' della Salute e della Scienza- Turin,
Department of Surgical Sciences - Nuclear Medicine,
Torino, Italy
3
A.O.U. Citta' della Salute e della Scienza- Turin,
Department of Medical Imaging - Medical Physics,
Torino, Italy
4
University of Turin A.O.U. Citta' della Salute e della
Scienza, Department of Surgical Sciences, Torino, Italy
5
University of Turin A.O.U. Citta' della Salute e della
Scienza, Department of Medical Sciences, Torino, Italy
6
A.O.U. Citta' della Salute e della Scienza- Turin,
Department of Oncology- Medical Oncology, Torino, Italy
7
University of Turin A.O.U. Citta' della Salute e della
Scienza, Department of Medical Imaging, Torino, Italy
8
A.O.U. Citta' della Salute e della Scienza- Turin,
Department of Oncology- Radiation Oncology, Torino,
Italy
Purpose or Objective
To test the proof of principle that irradiated volume of
pelvic active bone marrow (
ACT
BM), as detected by
18
FDG-
PET, may be a predictor of decreased blood cell nadirs in
anal cancer patients undergoing concurrent chemo-
radiation and to identify subregions within the pelvis
potentially more involved in the occurrence of acute
hematologic toxicity.
Material and Methods
Forty four patients submitted to IMRT and concurrent
chemotherapy were analyzed. Several bony structures
were defined: pelvic and lumbar-sacral (LSBM), lower
pelvis (LPBM) and iliac (IBM) bone marrow.
ACT
BM was
characterized employing
18
FDG-PET and defined as all
subregions within pelvic bone marrow having Standard
Uptake Values (SUVs) higher than SUV
mean
. All other regions
were defined as inactive BM (
INACT
BM) (Figure 1). On dose-
volume histograms, dosimetric parameters were taken.
Endpoints included white blood-cell-count (WBC),
absolute-neutrophil-count (ANC), hemoglobin (Hb) and
platelet (Plt) nadirs. Acute toxicity was scored according
to RTOG scoring scale. Generalized linear and logistic
regression models were used to find correlations between
dosimetric variables and blood cell toxicity.
Results
act
BM mean dose had a statistically significant correlation
with WBC (β=-1.338; 95%CI: -2.455/-0.221; p=0.020), ANC
(β=-1.651; 95%CI: -3.284/-0.183; p=0.048) and Plt (β=-
0.031; 95%CI: -0.057/-0.004; p=0.024) nadirs. On the
contrary, no correlation was found between
inact
BM mean
dose and any blood cell nadir (Table 1).
act
BM V
10
had a
significant correlation with WBC (β=-0.062; 95%CI: -
0.104/-0.021; p=0.004) and ANC (β=-0.038; 95%CI: -
0.067/-0.007; p=0.015) nadirs.
act
BM V
20
was significantly
correlated to WBC (β=-0.044; 95%CI: -0.080/-0.008;
p=0.017), ANC (β=-0.027; 95%CI: -0.052/-0.001; p=0.039)
and Plt (β=-1.570; 95%CI: -3.140/-0.002; p=0.050)
nadirs.
act
BM V
30
had a significant correlation with WBC
(β=-0.033; 95%CI: -0.064/-0.002; p=0.036) and Plt (β=-
1.720; 95%CI: -2.990/-0.450; p=0.010) nadirs.
act
BM V
40
was
significantly correlated to WBC (β=-1.490; 95%CI: -2.900/-
0.072; p=0.040) nadir. With respect to subregions within
the pelvis, WBC nadir was significantly correlated to
act
LSBM mean dose (β=-1.852; 95%CI: -3.205/-0.500;
p=0.009), V
10
(β=-2.153; 95%CI: -4.263/-0.721; p=0.002),
V
20
(β=-2.081; 95%CI: -4.880/-0.112; p=0.003), V
30
(β=-
1.971; 95%CI: -4.748/-0.090; p=0.023) and to
act
IBM V
10
(β=-0.073; 95%CI: -0.106/-0.023; p=0.016). ANC nadir
found to be significantly associated to
act
LSBM V
10
(β=-
1.878; 95%CI: -4.799/-0.643; p=0.025), V
20
(β=-1.765;
95%CI: -4.050/-0.613; p=0.030). No significant correlation
were found between dosimetric parameters and > G3
hematologic toxicity, even if borderline significance was
found for
act
LSBM mean dose and WBC nadir.