Abstract
Objectives : We
evaluated radiotherapy planning CT-based radiomics for predicting
clinical endpoints [tumor complete response (CR), 5-year overall
survival (OS), hypohemoglobin, and leucopenia] after
intensity-modulated radiation therapy (IMRT) in locally advanced
cervical cancer (LACC).
Design : A retrospectively study was performed between 2014 and
2017.
Setting : Xiangya hospital of Central South University,
Changsha, Hunan, China.
Population : A total of 257 LACC patients were treated with
IMRT.
Methods : Patients were allocated into the training/validation
sets (3:1 ratio) using proportional random sampling, resulting in the
same proportion of groups in the two sets.
Main outcomes and measures : The primary outcomes were the
treatment response and hematologic toxicities caused by radiotherapy. We
extracted 254 radiomic features from each of the gross target volume
(GTV), pelvis, and sacral vertebrae in planning CT images. The
sequentially backward elimination support vector machine algorithm was
used for feature selection and endpoint prediction. Model performance
was evaluated using area under the curve (AUC).
Results : A combination of 10 clinicopathological parameters and
34 radiomic features achieved the best performance for predicting CR
[validation balanced accuracy: 80.79%]. For OS, 54 radiomic
features showed good prediction accuracy [validation balanced
accuracy: 85.75%], and the threshold value of their scores can
stratify patients into the low-risk and high-risk groups
(P<0.001). The clinical and radiomic models were also
predictive of hypohemoglobin and severe
leucopenia [validation balanced
accuracies: 70.96% and 69.93%].
Conclusion : This study demonstrated that combining
clinicopathological parameters with CT-based radiomics had good
predictive value for treatment outcomes and hematologic toxicities to
radiotherapy in LACC. The prediction of clinical endpoints prior to
radiotherapy may assist the radiation therapists to select the optimal
therapeutic strategy with the minimal toxicity and best curative effect.
Funding :
Keywords : Radiation Oncology; Treatment Planning CT, Radiomics;
Treatment Outcome; Locally Advanced Cervical Cancer