PRO.M.E.THE.O.: the development of an italian multicentric OS predictive model in oropharynx cancer in Radiotherapy and Oncology
2020
AOU Città della Salute di Torino
Tipo pubblicazione
Conference Abstract
Autori/Collaboratori (18)Vedi tutti...
Miccichè F
Fondazione Policlinico Universitario A. Gemelli IRCCS, UOC di Radioterapia Oncologica- Dipartimento Diagnostica per Immagini- Radioterapia Oncologica ed Ematologia, Rome, Italy
Chiloiro G
Fondazione Policlinico Universitario A. Gemelli IRCCS, UOC di Radioterapia Oncologica- Dipartimento Diagnostica per Immagini- Radioterapia Oncologica ed Ematologia, Rome, Italy
Longo S
Università Cattolica del Sacro Cuore, Radiotherapy Oncology, Rome, Italy
et alii...
Abstract
Purpose or Objective In the last decade, remarkable advances in cancer care pose new challenges, leading clinical practice towards personalized medicine to understand the different behaviour of the same tumour in different patients. Aim of this study is developing a predictive model of overall survival (OS) for patients (pts) with oropharynx cancer treated with radiotherapy (RT) and Cetuximab (Cet), coming from PRO.M.E.THE.O.(PRedictiOn Models in Ent cancer for anti-EGFR based THErapy Optimization) italian multicentric dataset Material and Methods Pts with oropharynx cancer treated with RT-Cet coming from 6 centres (Cuneo, Firenze, Genova, Modena, Roma, Torino), were considered for this analysis.We considered different variables: age at diagnosis, gender, PS ECOG, TNM stage, total dose at CTV High Risk, fractionation, overall treatment time, RT interruption days.Follow-up times (2, 3, and 5 years) for the OS were used as the model outcome.Pts were randomly divided into two different groups: group A (training set) corresponding to 80% of dataset and group B (validation set) representing the remaining 20%.A binary logistic regression model was trained on group A with stepwise feature selection based on Akaike's information criterion (AIC). The model was then performed on testing set. ROC- AUC (Receiver Operating Characteristics- Area Under the Curve) and confusion matrix statistics at 5 threshold were used as performance criteria. Nomograms were performed at 2, 3 and 5 years. Results Of 218 pts collected, 183 (84%) were selected and analysed. OS predictive models were developed and represented by nomograms at 2,3 and 5 years. Covariates that show negative impact on 2y OS were positive nodal status, PS ECOG >1, Age, dose < 66Gy and interruption days of radiotherapy >3 days. ROC AUC for training set and testing set for OS at 2years was 0.787 (figure 1) and 0.763 (figure2)respectively Conclusion A large database allows the creation of an OS prediction model in oro
Se sei accreditato in BVS-P effettua l'accesso per utilizzare i nostri servizi.
DOI : 10.1016/S0167-8140(21)00859-8
Keywords
cetuximab; epidermal growth factor receptor antibody; adult; area under the curve; cancer radiotherapy; cancer staging; cancer survival; clinical trial; conference abstract; controlled study; drug therapy; feature selection; female; follow up; fractionation; gender; human; major clinical study; male; multicenter study; nomogram; oropharynx cancer; overall survival; prediction; radiotherapy; randomized controlled trial; receiver operating characteristic;