ESTRO 37 Abstract book

ESTRO 37

S359

Material and Methods This prospective cohort study included 186 head and neck cancer patients receiving primary (CH)RT and a complete dysphagia assessment (VF, SWAL-QOL and QLQ-H&N35 patient-rated and DYSPHAGIA RTOG, ASPIRATION CTCAEv4 physician-rated evaluation) at baseline and 6 months after treatment. Multiple imputation was used to account for missing data. The primary endpoint was penetration/aspiration 6 months after completing (CH)RT, scored on VF (PAS6), defined as a Penetration Aspiration Scale (PAS) score ≥ 3. From all baseline features, a set of candidate variables was selected, based on plausible relevance in relation to PAS6. For the selection of relevant patient- and physician-rated complaints, multivariate general linear model (MANCOVA) and logistic regression were used respectively, with correction for relevant covariates. Multivariable logistic regression with stepwise backward elimination was used to develop a model to predict PAS6. Results The prevalence of PAS6 was 44% (82 patients). Five patient-rated, aspiration-specific complaints were significantly associated with PAS. These complaints were included in the set of baseline candidate variables for the prediction model. There was no significant correlation between DYSPHAGIA RTOG /ASPIRATION CTCAEv4 and PAS and these parameters were not considered as candidate variables. (Fig.1) Multivariable analysis identified 3 independent prognostic factors for PAS6, including choking when drinking at baseline, stage T3-T4 and bilateral neck RT. After internal validation using bootstrapping the model performance was good (Nagelkerke R 2 : 0,22 and ROC- AUC: 0.73) (Fig 2.) and slightly improved after adding baseline PAS as candidate variable (ROC-AUC 0.74).

Figure1: Cox proportional hazards regression. (shown in graphic)

Conclusion Cervical lymph node group V is likely an im portant added prognostic factor to standard TNM staging. Further validation is warranted for confirmation and revision of TNM staging. PO-0704 NTCP model for aspiration after (chemo) radiation as a selection tool for baseline videofluoroscopy. A. Gawryszuk 1 , A. Van der Schaaf 1 , J.G.M. Vemer-van den Hoek 1 , H.P. Bijl 1 , J.A. Langendijk 1 1 University Medical Center Groningen, Radiation Oncology, Groningen, The Netherlands Purpose or Objective (Chemo) radiation (CH)RT is frequently used in head and neck cancer due to its non-invasive character permitting organ preservation. Nevertheless, larynx function is not always preserved and some patients require salvage surgery for dysfunctional larynx causing aspiration. In clinical practice, it is difficult to identify patients at risk of aspiration after (CH)RT due to its, often, silent nature and because not all patient- and physician-rated complaints (subjective measures) are consistent with videofluoroscopy (VF) findings (objective measures). The purposes of this study were (1) to identify subjective dysphagia measures associated with objectively (based on VF) scored aspiration and (2) to develop a prediction model, discriminating patients with a priori high risk of aspiration after (CH)RT, based on baseline characteristics.