PD-0448: Are advanced quantitative CT imaging features associated with survival in HNSCC patients?

E. Rios Velazquez, R. Leijenaar, F. Hoebers, J. Straetmans, B. Kremer, H. Aerts, P. Lambin
2013 Radiotherapy and Oncology  
Purpose/Objective: Curative (chemo-) radiation for head and neck cancer (HNC) may result in severe acute and late side effects, including tube feeding dependence. The purpose of this multicenter prospective cohort study was to develop a multivariate prediction model for tube feeding dependence 6 months (TUBE M6 ) after curative primary RT (radiotherapy), BioRT (radiotherapy + cetuximab) or CHRT (radiotherapy + chemotherapy) for head and neck cancer based on pre-treatment and treatment
more » ... stics (including dose volume parameters) that can be used to select patients for radiotherapy treatment plan optimization. Materials and Methods: The study included 416 patients with HNC. In all patients, TUBE M6 was scored prospectively in a standardized followup program. To design the prediction model, the penalized learning method LASSO was used, with TUBE M6 as the primary endpoint. Results: The incidence of TUBE M6 was 9.9% (41 out of 416 patients). Using the LASSO-parameter selection procedure, the multivariate model with the best model performance was selected, consisting of the following variables: male sex, advanced T-stage (T3-T4), moderate and severe weight loss at baseline, baseline CTCAE dysphagia-score, accelerated RT, CHRT, BioRT, the mean dose to the superior and inferior pharyngeal constrictor muscle (PCM), mean dose to the contralateral parotid gland and the mean dose to the cricopharyngeal muscle (CM) (see :Table) . Model performance at internal validation in terms of area under the curve was 0.90 in double cross validation, while the prediction model scored excellent using other performance measures. Of the four dose-volume parameters, the mean dose to the PCM superior was most important. In the graph (see: Figure) an example of the predicted NTCP value for TUBE M6 as a function of the mean dose to the PCM superior is shown for a female patient with a T3-T4 tumor, with severe weight loss and dysphagia prior to the start of chemoradiotherapy. Conclusions: We developed a multivariate NTCP model for TUBE M6 after definitive (chemo) radiation, which can be used to predict TUBE M6. The four dosimetric variables in this prediction model can be used to optimize RT treatment planning aiming at prevention of tube feeding dependence.
doi:10.1016/s0167-8140(15)32754-7 fatcat:g5qer2p3qnf27nr3g3m63pxnhe